Add ARC-Bench: 55-topic autonomous-research benchmark across 5 domains

.gitattributes

@@ -1,6 +1,5 @@
 *.7z filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text
-*.avro filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
 *.ckpt filter=lfs diff=lfs merge=lfs -text
@@ -10,7 +9,6 @@
 *.joblib filter=lfs diff=lfs merge=lfs -text
 *.lfs.* filter=lfs diff=lfs merge=lfs -text
 *.lz4 filter=lfs diff=lfs merge=lfs -text
-*.mds filter=lfs diff=lfs merge=lfs -text
 *.mlmodel filter=lfs diff=lfs merge=lfs -text
 *.model filter=lfs diff=lfs merge=lfs -text
 *.msgpack filter=lfs diff=lfs merge=lfs -text
@@ -35,7 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.xz filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
 # Audio files - uncompressed
 *.pcm filter=lfs diff=lfs merge=lfs -text
 *.sam filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,4 @@
+__pycache__/
+*.pyc
+.DS_Store
+.ipynb_checkpoints/

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Aiming Lab
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -0,0 +1,183 @@
+---
+license: mit
+language:
+- en
+pretty_name: ARC-Bench
+size_categories:
+- n<1K
+task_categories:
+- other
+tags:
+- autonomous-research
+- ai-agents
+- llm-agents
+- scientific-discovery
+- benchmark
+- machine-learning
+- high-energy-physics
+- quantum-computing
+- systems-biology
+- statistics
+configs:
+- config_name: default
+  data_files:
+  - split: test
+    path: data/arc_bench.jsonl
+- config_name: ml
+  data_files:
+  - split: test
+    path: data/ml.jsonl
+- config_name: physics
+  data_files:
+  - split: test
+    path: data/physics.jsonl
+- config_name: quantum
+  data_files:
+  - split: test
+    path: data/quantum.jsonl
+- config_name: biology
+  data_files:
+  - split: test
+    path: data/biology.jsonl
+- config_name: statistics
+  data_files:
+  - split: test
+    path: data/statistics.jsonl
+---
+
+# ARC-Bench
+
+**ARC-Bench** is a **55-topic, open-ended autonomous-research benchmark** spanning
+five scientific domains. Each topic is not a fixed-input/fixed-output task — it is a
+*research question* plus a structured briefing. A research agent (or a human) must
+take a topic from question → experiment design → code → measurements → claims →
+write-up, and the deliverable is graded against a weighted, multi-criteria rubric.
+
+| Domain | Count | IDs | Typical execution |
+|---|---|---|---|
+| Machine learning | 25 | `ML01`–`ML25` | CPU, `numpy`/`scipy`/`sklearn`/`statsmodels` |
+| High-energy physics | 10 | `P01`–`P10` | Lagrangian → MadGraph MC → analysis → figure (paper reproduction) |
+| Quantum | 10 | `Q01`–`Q10` | CPU, Qiskit 2.x statevector / VQE / QML |
+| Systems biology | 7 | `B01`–`B07` | Constraint-based modelling (COBRApy / BiGG) |
+| Statistics | 3 | `S01`–`S03` | Simulation studies (`numpy`/`scipy`/`statsmodels`) |
+
+The ML, quantum, and statistics topics are **open research questions** (the agent
+designs the experiment); the physics topics are **published-paper reproductions**
+(each scoped to a specific reference figure); the biology topics are
+**constraint-based metabolic-modelling** studies.
+
+## Load it
+
+```python
+from datasets import load_dataset
+
+# all 55 topics
+ds = load_dataset("AIMING-Lab-UNC/ARC-Bench", split="test")
+
+# a single domain subset
+ml = load_dataset("AIMING-Lab-UNC/ARC-Bench", "ml", split="test")
+
+row = ds[0]
+print(row["id"], row["title"])
+print(row["metric_key"], row["metric_direction"])
+```
+
+Available config names: `default` (all 55), `ml`, `physics`, `quantum`,
+`biology`, `statistics`.
+
+## Schema
+
+Each row describes one benchmark topic. Deeply-nested / variable-shape fields are
+stored as **JSON-encoded strings** so the table schema is stable across all
+domains; parse them with `json.loads`.
+
+| Column | Type | Description |
+|---|---|---|
+| `id` | string | Topic id (`ML01`, `P03`, `Q07`, `B01`, `S02`, …) |
+| `domain` | string | One of `ml` / `physics` / `quantum` / `biology` / `statistics` |
+| `title` | string | Human-readable topic title |
+| `topic` | string | One-line topic statement (from the domain registry) |
+| `domains` | list[string] | Subfield tags (e.g. `["machine-learning","calibration"]`) |
+| `arxiv_id` | string \| null | Source paper (physics reproductions; null for open questions) |
+| `venue` | string | Benchmark venue label |
+| `metric_key` | string | Headline metric name |
+| `metric_direction` | string | `maximize` / `minimize` / `match_reference` |
+| `gpu_required` | bool | Whether a GPU is needed (all topics are CPU-friendly → `false`) |
+| `est_wall_clock_sec` | int | Rough single-run wall-clock budget |
+| `synthesis` | string | The research briefing: background + what a credible study includes |
+| `num_hypotheses` | int | Number of pre-registered hypotheses |
+| `hypotheses` | string (JSON) | List of `{id, statement, measurable}` |
+| `experiment_design` | string (JSON) | `research_question`, `conditions`, `baselines`, `metrics`, `datasets`, `compute_requirements` |
+| `requirements` | string (JSON) | Agent-mode pass/fail gating items (physics + biology; `""` otherwise) |
+| `rubric` | string (JSON) | Hierarchical weighted scoring rubric (code / execution / results buckets) |
+| `rubric_num_leaves` | int | Number of leaf criteria in the rubric |
+| `manifest_file` | string | Path to the raw manifest inside this repo (`tasks/…`) |
+| `rubric_file` | string | Path to the raw rubric inside this repo (`tasks/…`) |
+
+### Raw inputs
+
+The flattened `data/*.jsonl` is convenient for `load_dataset`. The authoritative,
+human-readable benchmark inputs are also shipped verbatim under `tasks/`:
+
+```
+tasks/
+├── meta_paper_quality.json         # shared paper-quality meta-rubric (manual grading)
+└── <domain>/
+    ├── topics.yaml                 # the domain topic registry
+    ├── manifests/<ID>.yaml         # full per-topic briefing
+    └── rubrics/<ID>.json           # weighted scoring rubric
+```
+
+## How a topic is scored
+
+Each topic carries a hierarchical rubric. For ML / quantum / statistics it has
+three buckets — **Code Development**, **Code Execution**, **Result Analysis**
+(weighted roughly 25 : 25 : 50). Physics and biology add a fourth
+**Reproducibility** bucket. Leaf criteria are graded on scientific substance and
+directional correctness of the evidence, not on rigid threshold matching (see each
+rubric's `judging_note`).
+
+A second, optional layer — `tasks/meta_paper_quality.json` — grades the **paper
+output** (writing, code orchestration, figure quality, factual accuracy) and is
+intended for manual / vision-equipped grading rather than fast automated scoring.
+
+## Intended use
+
+- Evaluating autonomous-research / AI-scientist agents end-to-end.
+- Studying agent behavior across heterogeneous scientific domains with a *single*
+  task format.
+- As a stimulus set for human-in-the-loop or framework-comparison studies.
+
+The runner harness, baseline adapters, and judges are **not** part of this dataset;
+they live in the source repository (link below).
+
+## Attribution
+
+The benchmark glue (manifests, rubrics, registries) is the authors' own work. Some
+domain pipelines are driven by **external Claude-Code agents**, which should be
+credited when reporting domain results:
+
+| Topic family | External agent | Upstream |
+|---|---|---|
+| `P01`–`P10` (HEP) | ColliderAgent | <https://github.com/HET-AGI/ColliderAgent> |
+| `B01`–`B07` (metabolic) | Biology-Agent | constraint-based modelling pipeline |
+
+## Links
+
+- **Code / harness:** <https://github.com/aiming-lab/AutoResearchClaw>
+- **Paper page:** <https://huggingface.co/papers/2605.20025>
+
+## Citation
+
+```bibtex
+@misc{arcbench2026,
+  title  = {ARC-Bench: An Open-Ended Autonomous-Research Benchmark Across Five Scientific Domains},
+  author = {Qiu, Shi and others},
+  year   = {2026},
+  note   = {https://huggingface.co/datasets/AIMING-Lab-UNC/ARC-Bench}
+}
+```
+
+## License
+
+Released under the [MIT License](LICENSE).

data/biology.jsonl

@@ -0,0 +1,7 @@
+{"id": "B01", "domain": "biology", "title": "E. coli succinate-production strain optimization via constraint-based knockout screen", "topic": "E. coli succinate-production strain optimization via constraint-based knockout screen on iAF1260", "domains": ["systems-biology", "metabolic-engineering", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 1800, "synthesis": "Succinate is a top-12 platform chemical and a recurring case study in\nmetabolic engineering. The standard genome-scale model E. coli iAF1260\n(2382 reactions, 1668 metabolites, 1261 genes) provides a well-validated\ntestbed for in-silico strain design: from a fixed glucose-minimal medium\none can predict (a) the maximum aerobic biomass growth rate, (b) the\nmaximum theoretical succinate secretion flux once growth is held at a feasible\nfraction of optimum, (c) the set of single- and double-gene knockouts\nthat decouple biomass from succinate secretion enough to push the strain\ntoward a high-secretion non-growth-coupled phenotype.\n\nA credible CPU-scale study of this topic (a) loads iAF1260 from BIGG and\nvalidates aerobic biomass > 0.7 1/h on glucose-minimal medium, (b) runs\nparsimonious FBA (pFBA) at the wild-type optimum and reports the central\ncarbon flux distribution, (c) sweeps the production envelope (biomass vs\nsuccinate secretion) using flux variability analysis (FVA) on the\nsuccinate exchange across a grid of biomass-fraction-of-optimum values,\n(d) runs a single-gene knockout screen (≤100 candidate genes drawn from\ncentral carbon metabolism, fermentation, and TCA cycle) and ranks each KO\nby post-KO succinate secretion flux at 50% of WT growth, (e) reports the top three\nKO strains with mechanistic explanation tied to the network topology.\n\nThe research question is: *which single-gene knockouts of central carbon\nmetabolism in E. coli iAF1260 produce the largest predicted succinate\nsecretion flux while preserving at least 50% of wild-type biomass growth, and what\nis the mechanistic role of each in shifting flux toward succinate?*", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"Wild-type aerobic E. coli iAF1260 on glucose-minimal medium predicts a biomass growth rate within ±10% of 0.736 1/h (the published BIGG iAF1260 reference value).\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"The biomass-vs-succinate production envelope, computed via FVA at biomass fractions from 0.1 to 1.0 of WT optimum, is monotonically non-increasing in succinate as biomass approaches the WT optimum (i.e., succinate is growth-competing under aerobic glucose conditions).\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"At least 3 of the top-5 single-gene knockouts identified by the screen as maximizing succinate secretion flux at 50% WT-growth disrupt canonical competing by-product or respiration pathways described in the metabolic-engineering literature (e.g., pyruvate-formate-lyase pflB, lactate dehydrogenase ldhA, alcohol dehydrogenase adhE, acetate kinase ackA, phosphate acetyltransferase pta, or their isozymes), or are accompanied by a mechanistic explanation for why the knockout redirects flux toward succinate under the chosen medium.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Which single-gene knockouts of central carbon metabolism in E. coli iAF1260 produce the largest predicted succinate secretion flux while preserving at least 50% of wild-type biomass growth, and what is the mechanistic role of each in shifting flux toward succinate?\", \"conditions\": [{\"name\": \"wt_aerobic_glucose\", \"description\": \"Wild-type iAF1260 on glucose-minimal medium with O2 unconstrained. Reference condition.\"}, {\"name\": \"wt_aerobic_glucose_pfba\", \"description\": \"Same as wt_aerobic_glucose but with pFBA (parsimonious FBA) for unique flux distribution.\"}, {\"name\": \"production_envelope_succinate\", \"description\": \"FVA on succinate exchange across biomass-fraction-of-optimum grid {0.1, 0.2, ..., 1.0}.\"}, {\"name\": \"single_ko_screen_central_carbon\", \"description\": \"Single-gene knockouts on a curated set of ≤100 central-carbon / fermentation / TCA genes (glycolysis, PPP, TCA, anaplerotic, fermentation by-products); for each, FBA at fixed biomass = 0.5 × WT optimum; record succinate exchange flux.\"}], \"baselines\": [\"wt_aerobic_glucose (no perturbation) is the no-engineering baseline\", \"Published BIGG iAF1260 reference growth rate (0.736 1/h on glucose-minimal aerobic) — the model-validity gate\"], \"metrics\": [{\"name\": \"wt_growth_rate\", \"direction\": \"match_reference\", \"description\": \"Wild-type biomass exchange flux (1/h) on glucose-minimal aerobic; reference value 0.736.\"}, {\"name\": \"max_succinate_flux_mmol_gDW_h_at_zero_growth\", \"direction\": \"maximize\", \"description\": \"Succinate exchange flux (mmol/gDW/h) at biomass = 0 (theoretical maximum).\"}, {\"name\": \"succinate_flux_mmol_gDW_h_at_half_growth\", \"direction\": \"maximize\", \"description\": \"Succinate exchange flux at biomass = 0.5 × WT optimum, wild-type strain.\"}, {\"name\": \"best_ko_succinate_flux_mmol_gDW_h\", \"direction\": \"maximize\", \"description\": \"Maximum succinate exchange flux over the single-KO screen at biomass = 0.5 × WT optimum.\"}], \"datasets\": [{\"name\": \"iAF1260\", \"source\": \"BIGG database (http://bigg.ucsd.edu/models/iAF1260) via cobra.io.load_model('iAF1260')\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 1800}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"A canonical results.json file exists at the workspace root with at least the keys: primary_metric (number), metrics (object of numeric keys), hypotheses (object with h1/h2/h3 entries each carrying a `supported` boolean), summary (non-empty string).\", \"must_pass\": true}, {\"id\": \"req_wt_growth_value\", \"type\": \"numeric\", \"description\": \"results.json metrics MUST contain a numeric value for `wt_growth_observed_1_per_h` (or an equivalent wt_growth_*_1_per_h key) that is non-null and finite. The wild-type growth rate must be reported, otherwise H1 cannot be evaluated.\", \"must_pass\": true}, {\"id\": \"req_top5_ko\", \"type\": \"discussion\", \"description\": \"results.json structured_results MUST contain a list of the top 5 ranked single-gene knockouts with at least {gene_id, gene_name, ko_succinate_flux_mmol_gDW_h, is_canonical} fields each. The list need not include the textbook canonical genes — but it MUST exist and identify at least 5 candidates.\", \"must_pass\": true}, {\"id\": \"req_envelope_figure\", \"type\": \"artifact\", \"description\": \"A production-envelope figure (biomass-vs-succinate, or analogous) must exist under figures/ or analysis/ in PDF or PNG format with axes labeled and units shown. Either a single multi-curve figure or per-condition figures both qualify.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"Top-3 KO targets are discussed mechanistically — naming the pathway each KO disables and a one-sentence rationale for why the disruption redirects flux to succinate.  Acceptable inside summary, structured_results, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and any RNG seeds used (e.g. for sampling) are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B01\", \"requirements\": \"A credible constraint-based experiment on E. coli iAF1260 that (a) loads the BIGG model and validates wild-type growth, (b) computes a flux distribution under the standard objective, (c) characterises the biomass-vs-succinate production envelope, (d) runs a single-gene knockout screen targeted at central carbon metabolism, and (e) ties the top-ranked knockouts to the network mechanism. Partial but well-motivated evidence deserves partial credit; rigid wording or naming should not penalize a substantively correct experiment.\", \"judging_note\": \"Score on biological substance and quantitative correctness, not exact threshold satisfaction. Reproducing canonical FBA numbers (WT growth ~0.7, well-known knockouts like pflB/ldhA/adhE/ackA-pta) is strong evidence; novel-but-coherent design choices are acceptable.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b01-code\", \"requirements\": \"The constraint-based pipeline is implemented with COBRApy and a BIGG genome-scale model.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b01-code-model\", \"requirements\": \"The submission loads E. coli iAF1260 (or a comparable validated genome-scale metabolic model) from BIGG via COBRApy, with the medium constraints (glucose uptake, O2 availability) explicitly set and the biomass objective explicitly named.\", \"weight\": 8.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b01-code-fba\", \"requirements\": \"FBA and pFBA are invoked correctly to obtain wild-type growth and a unique flux distribution; output flux table covers exchange + central-carbon reactions at minimum.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b01-code-envelope\", \"requirements\": \"The biomass-vs-succinate production envelope is computed by sweeping a fraction-of-optimum biomass constraint and running FVA (or two-step FBA) on the succinate exchange at each grid point.\", \"weight\": 4.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b01-code-koscreen\", \"requirements\": \"The single-gene knockout screen iterates over a curated central-carbon / fermentation / TCA gene set (≤100), applies model.genes.<g>.knock_out() (or equivalent reaction-bound deletion), and records succinate flux at fixed biomass = 0.5 × WT optimum.\", \"weight\": 4.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b01-exec\", \"requirements\": \"Execution produces the FBA / FVA / KO numbers needed to evaluate the hypotheses without crashing.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b01-exec-validity\", \"requirements\": \"Solver returns optimal status for the wild-type FBA and at least 90% of the KO conditions; no negative biomass or non-physical fluxes (succinate secretion negative under aerobic glucose, etc.) appear without explanation.\", \"weight\": 8.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b01-exec-physics\", \"requirements\": \"Mass balance, charge balance, and biomass-positive checks pass for the loaded model; this is the biology-validity gate analogous to physics-validity (gauge invariance, unitarity) in the physics rubric.\", \"weight\": 8.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b01-exec-results\", \"requirements\": \"A machine-readable results artifact (results.json or simulations/*.csv) records WT growth rate, the production-envelope grid (biomass fraction → succinate secretion flux in mmol/gDW/h), and a per-KO succinate-flux table. If yield is reported, define it separately as succinate secretion flux divided by glucose uptake flux.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b01-results\", \"requirements\": \"The results analysis addresses the three hypotheses with quantitative evidence and a clear narrative tied to network mechanism.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b01-result-h1-quant\", \"requirements\": \"Quantitative test of H1: predicted WT aerobic growth rate is reported and compared to the published BIGG reference 0.736 1/h. Score 100% if relative error < 3%, 67% if < 10%, 33% if < 30%, 0% otherwise.\", \"weight\": 12.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b01-result-h2-quant\", \"requirements\": \"Quantitative test of H2: production envelope shows succinate is growth-competing — succinate flux is non-increasing as biomass approaches WT optimum. Score 100% if monotonic over ≥8 of 10 grid points, 67% if ≥6 of 10, 33% if ≥4 of 10.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b01-result-h3-quant\", \"requirements\": \"Quantitative test of H3: among the top-5 KO ranked by succinate secretion flux at 0.5 × WT growth, count how many disrupt canonical competing by-product or respiration pathways for succinate-overproduction logic, such as {pflB, ldhA, adhE, ackA, pta, acetate/ethanol/lactate branch competitors, oxygen-respiration competitors}, or include a specific mechanistic explanation for redirecting flux toward succinate under the chosen medium. Score 100% if ≥3 of 5 meet this canonical-or-mechanistic criterion, 67% if 2 of 5, 33% if 1 of 5, 0% otherwise.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b01-result-figure\", \"requirements\": \"At least one publication-quality figure is produced — typically the production envelope (biomass on x, succinate secretion flux on y, with WT and best KO overlaid) and/or the KO essentiality / flux heatmap. Axes labeled with units (1/h, mmol/gDW/h), legend present.\", \"weight\": 8.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b01-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, names the top knockouts with their gene/enzyme identity, and gives a one-paragraph mechanistic interpretation (which competing pathways the KOs disable, why this redirects flux to succinate).\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b01-repro\", \"requirements\": \"Reproducibility: the model, run cards, and seeds are checked into artifacts so a fresh clone can rerun and obtain matching numbers.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b01-repro-model\", \"requirements\": \"The loaded GSMM is either persisted (models/iAF1260.json) or pinned by version + BIGG URL in the writeup; a downstream user can locate the exact model object used.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b01-repro-runcards\", \"requirements\": \"The medium definition (exchange bounds), objective function, FVA fraction-of-optimum, and KO gene list are saved to a config / params file (not only inline in code).\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b01-repro-seeds\", \"requirements\": \"Solver settings (LP backend name + version, tolerance) are recorded; if any sampling step is used, the RNG seed is logged. A second run reproduces the central WT growth rate to within solver tolerance.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B01.yaml", "rubric_file": "tasks/biology/rubrics/B01.json"}
+{"id": "B02", "domain": "biology", "title": "E. coli acetate overflow metabolism phase map under glucose and oxygen limitation", "topic": "E. coli acetate overflow metabolism phase map under glucose and oxygen limitation", "domains": ["systems-biology", "constraint-based-modelling", "phase-plane-analysis"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 1800, "synthesis": "Overflow acetate secretion in E. coli is a canonical constraint-based\nmodelling phenotype. A credible CPU-scale study loads a validated E. coli\ngenome-scale model, fixes a glucose-minimal medium, sweeps glucose and oxygen\nuptake bounds, computes growth and major secretion products, and identifies\nthe transition from respiratory growth to overflow/fermentative by-product\nsecretion.\n\nThe research question is: under which glucose and oxygen uptake regimes does\nE. coli switch from primarily respiratory growth to acetate-secreting overflow\nmetabolism, and how robust is that regime boundary across FBA and pFBA?", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"The selected E. coli BIGG model grows on aerobic glucose-minimal medium with a positive biomass objective and no infeasible solver status.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"A two-dimensional glucose-vs-oxygen phase plane contains distinct regimes with high oxygen/low acetate and low oxygen/high acetate secretion.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"pFBA reduces total internal flux while preserving the same growth-rate regime boundaries to within a small tolerance.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Under which glucose and oxygen uptake regimes does E. coli switch from respiratory growth to acetate overflow?\", \"conditions\": [{\"name\": \"aerobic_glucose_reference\", \"description\": \"Wild-type E. coli iJO1366 or iAF1260 on glucose-minimal medium with oxygen available.\"}, {\"name\": \"glucose_oxygen_phase_plane\", \"description\": \"Grid over glucose uptake and oxygen uptake bounds; compute biomass, acetate, ethanol, lactate, succinate, and CO2 exchange fluxes.\"}, {\"name\": \"pfba_phase_plane_check\", \"description\": \"Run pFBA on the same grid or a representative subset and compare growth plus secretion regimes.\"}], \"baselines\": [\"High-oxygen glucose condition is the respiratory baseline.\", \"Zero-oxygen glucose condition is the fermentative baseline.\"], \"metrics\": [{\"name\": \"growth_rate_1_per_h\", \"direction\": \"maximize\", \"description\": \"Biomass objective flux for each nutrient grid point.\"}, {\"name\": \"acetate_flux_mmol_gDW_h\", \"direction\": \"characterize\", \"description\": \"Acetate exchange secretion flux.\"}, {\"name\": \"overflow_boundary_o2_bound\", \"direction\": \"estimate\", \"description\": \"Approximate oxygen bound below which acetate secretion exceeds a chosen threshold.\"}, {\"name\": \"pfba_total_flux_norm\", \"direction\": \"minimize\", \"description\": \"Total absolute flux under pFBA.\"}], \"datasets\": [{\"name\": \"E. coli GSMM\", \"source\": \"BIGG iJO1366 or iAF1260 via cobra.io.load_model\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 1800}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"A results.json file exists with primary_metric, metrics, hypotheses, summary, and structured_results.\", \"must_pass\": true}, {\"id\": \"req_phase_plane_table\", \"type\": \"artifact\", \"description\": \"A machine-readable phase-plane table exists with glucose_bound, oxygen_bound, growth_rate, and acetate_flux columns.\", \"must_pass\": true}, {\"id\": \"req_phase_plane_figure\", \"type\": \"artifact\", \"description\": \"A PNG or PDF heatmap/phase-plane figure exists with glucose and oxygen axes and units.\", \"must_pass\": true}, {\"id\": \"req_aerobic_ref_growth\", \"type\": \"numeric\", \"description\": \"results.json metrics MUST contain a numeric value for `aerobic_glucose_growth_rate_1_per_h` (or an equivalent aerobic_growth_*_1_per_h key) that is non-null, finite, and positive. The aerobic reference growth rate must be reported, otherwise H1 cannot be evaluated.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"The overflow boundary and its mechanistic basis are discussed — why E. coli switches to acetate-secreting overflow at the identified nutrient threshold. Acceptable inside summary, phase_plane_interpretation, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B02\", \"requirements\": \"A credible constraint-based E. coli acetate-overflow phase-plane study that loads a validated BIGG model, sets glucose and oxygen medium bounds, sweeps a two-dimensional nutrient grid, reports growth and by-product secretion, and interprets the acetate overflow boundary.\", \"judging_note\": \"Score biological substance and quantitative reproducibility over exact thresholds. iJO1366, iAF1260, or a comparable curated E. coli GSMM are acceptable.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b02-code\", \"requirements\": \"The COBRApy implementation defines the model, medium, phase-plane grid, FBA/pFBA runs, and secretion extraction.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b02-code-model\", \"requirements\": \"Loads a validated E. coli BIGG model, explicitly sets glucose and oxygen exchange bounds, and names the biomass objective.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b02-code-grid\", \"requirements\": \"Constructs a two-dimensional glucose-by-oxygen grid with at least 8 values per axis and restores model state between conditions.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b02-code-products\", \"requirements\": \"Extracts biomass plus acetate, ethanol, lactate, succinate, CO2, and O2/glucose exchange fluxes where present.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b02-code-pfba\", \"requirements\": \"Runs pFBA on the full grid or a justified representative subset and records total absolute flux.\", \"weight\": 4, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b02-exec\", \"requirements\": \"Execution produces stable phase-plane artifacts without infeasible or missing-condition failures dominating the study.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b02-exec-status\", \"requirements\": \"At least 90% of grid points solve to optimal or biologically expected zero-growth status with explicit handling.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b02-exec-physics\", \"requirements\": \"Mass-balance check passes for the loaded model; no non-physical secretion fluxes appear under aerobic conditions without explicit justification (e.g. acetate secretion is zero under high-oxygen/low-glucose conditions in the reference FBA); biomass flux is positive under aerobic glucose reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b02-exec-artifacts\", \"requirements\": \"Writes results.json and a CSV/TSV phase-plane table with nutrient bounds, growth, and secretion fluxes.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b02-results\", \"requirements\": \"The analysis identifies and explains acetate-overflow regimes with figures and quantitative criteria.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b02-result-h1\", \"requirements\": \"Reports reference aerobic glucose growth and confirms model feasibility and positive biomass production. Score 100% if a positive finite biomass flux is reported for aerobic glucose with optimal solver status; 67% if growth is positive but solver status is not verified; 33% if only feasibility is stated without a numeric growth value; 0% otherwise.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b02-result-h2\", \"requirements\": \"Quantifies an acetate secretion boundary or threshold across oxygen/glucose regimes and distinguishes respiratory vs overflow conditions. Score 100% if an acetate-overflow boundary is quantified (O2 or glucose uptake threshold stated) with ≥2 distinct secretion regimes visible in the phase-plane table; 67% if overflow is identified without a quantified boundary; 33% if only one secretion regime is described; 0% otherwise.\", \"weight\": 12, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b02-result-h3\", \"requirements\": \"Compares FBA and pFBA growth/secretion regimes and explains where they agree or differ. Score 100% if FBA and pFBA growth rates are numerically compared and total flux norms are reported for both; 67% if only growth rates are compared without flux-norm comparison; 33% if pFBA was run but not compared to FBA; 0% otherwise.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b02-result-figure\", \"requirements\": \"Produces at least one labelled heatmap or contour plot for growth and acetate secretion with units.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b02-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, characterises the acetate-overflow regime boundary with quantitative support, and gives a one-paragraph mechanistic interpretation of why E. coli shifts from respiratory to acetate-secreting overflow metabolism at the identified nutrient threshold.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b02-repro\", \"requirements\": \"The model ID, medium, grid values, solver, and thresholds are recorded for rerun reproducibility.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b02-repro-model\", \"requirements\": \"The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object used.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b02-repro-config\", \"requirements\": \"Saves model ID, objective reaction, nutrient grid, and secretion threshold in config or results metadata.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b02-repro-solver\", \"requirements\": \"Records solver backend, tolerance, COBRApy version, and rerun consistency for the aerobic reference point.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B02.yaml", "rubric_file": "tasks/biology/rubrics/B02.json"}
+{"id": "B03", "domain": "biology", "title": "Anaerobic E. coli lactate-overproduction knockout screen", "topic": "Anaerobic E. coli lactate-overproduction knockout screen", "domains": ["systems-biology", "metabolic-engineering", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 1800, "synthesis": "Lactate production under anaerobic glucose conditions is a tractable\nmetabolic-engineering task for constraint-based modelling. A credible study\nloads an E. coli BIGG model, validates anaerobic growth, computes WT lactate\nsecretion, screens central-carbon and fermentation knockouts, and ranks\ninterventions that increase lactate secretion while preserving growth.", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"The model supports positive anaerobic glucose growth after oxygen uptake is closed.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"The lactate production envelope shows a tradeoff between biomass growth and maximum lactate secretion.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Top lactate-improving knockouts disable competing ethanol, formate, acetate, or succinate by-product routes, or have a specific mechanistic explanation.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Which single-gene knockouts increase anaerobic lactate secretion in E. coli while preserving at least 30-50% of WT growth?\", \"conditions\": [{\"name\": \"wt_anaerobic_glucose\", \"description\": \"E. coli iJO1366 or iAF1260 on glucose-minimal medium with oxygen uptake closed.\"}, {\"name\": \"lactate_production_envelope\", \"description\": \"Sweep biomass fraction and optimize lactate secretion.\"}, {\"name\": \"single_ko_screen_fermentation\", \"description\": \"Screen central-carbon and fermentation genes; rank by lactate flux at fixed growth fraction.\"}], \"baselines\": [\"Wild-type anaerobic glucose lactate secretion.\", \"No-knockout lactate envelope.\"], \"metrics\": [{\"name\": \"anaerobic_wt_growth_1_per_h\", \"direction\": \"maximize\", \"description\": \"WT biomass flux with oxygen uptake set to zero.\"}, {\"name\": \"lactate_flux_mmol_gDW_h\", \"direction\": \"maximize\", \"description\": \"Lactate exchange secretion flux.\"}, {\"name\": \"lactate_yield_per_glucose\", \"direction\": \"maximize\", \"description\": \"Lactate secretion divided by glucose uptake.\"}, {\"name\": \"growth_fraction\", \"direction\": \"threshold\", \"description\": \"KO biomass relative to WT anaerobic growth.\"}], \"datasets\": [{\"name\": \"E. coli GSMM\", \"source\": \"BIGG iJO1366 or iAF1260 via cobra.io.load_model\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 1800}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"results.json contains primary_metric, metrics, hypotheses, summary, and top KO structured results.\", \"must_pass\": true}, {\"id\": \"req_anaerobic_medium\", \"type\": \"numeric\", \"description\": \"Oxygen uptake is explicitly closed and anaerobic WT growth is reported.\", \"must_pass\": true}, {\"id\": \"req_top5_ko\", \"type\": \"discussion\", \"description\": \"Top 5 lactate-ranked knockouts are reported with gene_id, gene_name, growth_fraction, lactate_flux, and mechanism/canonical flag.\", \"must_pass\": true}, {\"id\": \"req_envelope_figure\", \"type\": \"artifact\", \"description\": \"A biomass-vs-lactate production envelope figure exists with labelled axes and units.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"Top-3 lactate-improving KO targets are discussed mechanistically — naming the pathway each KO disables and a one-sentence rationale for why the disruption redirects anaerobic flux to lactate. Acceptable inside summary, structured_results, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B03\", \"requirements\": \"A credible anaerobic E. coli lactate-overproduction study using COBRApy that validates anaerobic growth, computes a lactate production envelope, screens a focused central-carbon/fermentation knockout set, and explains top candidates mechanistically.\", \"judging_note\": \"Accept iJO1366, iAF1260, or comparable curated E. coli models. Score coherent anaerobic setup and mechanistic KO interpretation over exact gene names.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b03-code\", \"requirements\": \"COBRApy code implements anaerobic medium, lactate objective/envelope, and knockout screening.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b03-code-medium\", \"requirements\": \"Loads an E. coli BIGG model, sets glucose uptake, closes oxygen uptake, and names biomass and lactate exchange reactions.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b03-code-envelope\", \"requirements\": \"Computes biomass-vs-lactate production envelope by constraining biomass fractions and optimizing or FVA-bounding lactate secretion.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b03-code-ko\", \"requirements\": \"Screens a focused set of no more than 100 central-carbon and fermentation genes with model state restored between knockouts.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b03-code-yield\", \"requirements\": \"Computes lactate yield as lactate secretion divided by absolute glucose uptake, separate from flux.\", \"weight\": 4, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b03-exec\", \"requirements\": \"Execution produces valid anaerobic WT, envelope, and KO tables.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b03-exec-validity\", \"requirements\": \"WT anaerobic FBA is optimal with positive biomass and at least 85% of KO simulations solve or are explicitly labelled infeasible/lethal.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b03-exec-physics\", \"requirements\": \"Mass-balance check passes for the loaded anaerobic model; no aerobic by-products (O2 uptake or CO2 via respiration) appear under strictly anaerobic conditions without explanation; biomass flux is positive under anaerobic glucose reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b03-exec-artifacts\", \"requirements\": \"Writes results.json plus machine-readable envelope and KO-ranking tables.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b03-results\", \"requirements\": \"Results quantify lactate production, growth tradeoffs, and KO mechanisms.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b03-result-h1\", \"requirements\": \"Reports anaerobic WT growth and confirms oxygen uptake is zero or bounded closed. Score 100% if positive anaerobic WT biomass flux is reported with O2 exchange confirmed ≤0 mmol/gDW/h; 67% if growth is positive but O2 status is not explicitly confirmed; 33% if anaerobic feasibility is stated without a numeric growth value; 0% otherwise.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b03-result-h2\", \"requirements\": \"Shows lactate secretion changes across biomass fractions and identifies whether production is growth-coupled, competing, or partially coupled. Score 100% if lactate secretion flux is non-decreasing as biomass fraction decreases across ≥8 of 10 envelope grid points; 67% if non-decreasing over ≥6 of 10 points; 33% if ≥4 of 10 points; 0% otherwise.\", \"weight\": 10, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b03-result-h3\", \"requirements\": \"Top 5 KO list includes growth fraction, lactate flux/yield, and pathway-level rationale for at least top 3 targets. Score 100% if ≥3 of the top-5 KOs disrupt canonical competing by-product pathways (acetate, ethanol, formate, or succinate routes) or include a specific mechanistic explanation; 67% if 2 of 5 meet this criterion; 33% if 1 of 5; 0% otherwise.\", \"weight\": 12, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b03-result-figure\", \"requirements\": \"Produces a labelled lactate envelope or KO ranking figure with units.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b03-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, names the top lactate-improving knockouts with their gene/enzyme identity, and gives a one-paragraph mechanistic interpretation (which competing pathways the KOs disable, why this redirects anaerobic flux to lactate).\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b03-repro\", \"requirements\": \"Medium, model, gene set, solver, and thresholds are reproducibly recorded.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b03-repro-model\", \"requirements\": \"The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce the anaerobic medium setup.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b03-repro-runcard\", \"requirements\": \"Saves anaerobic medium bounds, biomass threshold, KO gene set, and lactate reaction ID.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b03-repro-solver\", \"requirements\": \"Records COBRApy/solver versions and confirms rerun consistency for WT growth.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B03.yaml", "rubric_file": "tasks/biology/rubrics/B03.json"}
+{"id": "B04", "domain": "biology", "title": "S. cerevisiae ethanol yield under oxygen limitation and carbon-source changes", "topic": "S. cerevisiae ethanol yield under oxygen limitation and carbon-source changes", "domains": ["systems-biology", "metabolic-engineering", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 1800, "synthesis": "Yeast ethanol fermentation is a natural extension beyond E. coli while still\nstaying inside BIGG/COBRApy constraint-based modelling. A credible study\nloads iMM904 or a comparable yeast GSMM, validates growth on glucose, sweeps\noxygen availability, compares glucose and alternative carbon sources, and\nquantifies ethanol secretion and yield.", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"The yeast model grows on glucose-minimal medium and produces a feasible FBA solution.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"Ethanol secretion/yield increases as oxygen uptake is restricted relative to fully aerobic growth.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Carbon-source swaps produce distinct growth and ethanol-yield profiles, with glucose supporting stronger fermentation than at least one alternative carbon source.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"How do oxygen limitation and carbon source alter predicted ethanol secretion and yield in S. cerevisiae?\", \"conditions\": [{\"name\": \"yeast_glucose_reference\", \"description\": \"S. cerevisiae iMM904 or comparable model on glucose-minimal medium.\"}, {\"name\": \"oxygen_sweep\", \"description\": \"Sweep oxygen uptake from anaerobic/low oxygen to aerobic while glucose uptake is fixed.\"}, {\"name\": \"carbon_source_swap\", \"description\": \"Compare glucose, fructose, galactose, glycerol, and acetate when supported by the model.\"}], \"baselines\": [\"Aerobic glucose condition.\", \"Anaerobic or oxygen-limited glucose condition.\"], \"metrics\": [{\"name\": \"growth_rate_1_per_h\", \"direction\": \"maximize\", \"description\": \"Biomass flux.\"}, {\"name\": \"ethanol_flux_mmol_gDW_h\", \"direction\": \"maximize\", \"description\": \"Ethanol exchange secretion flux.\"}, {\"name\": \"ethanol_yield_per_carbon_uptake\", \"direction\": \"maximize\", \"description\": \"Ethanol secretion divided by substrate uptake.\"}], \"datasets\": [{\"name\": \"S. cerevisiae iMM904\", \"source\": \"BIGG iMM904 via cobra.io.load_model if available, or documented local SBML/JSON yeast model.\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 1800}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"results.json contains yeast model ID, metrics, hypotheses, and structured oxygen/carbon-source results.\", \"must_pass\": true}, {\"id\": \"req_oxygen_sweep\", \"type\": \"artifact\", \"description\": \"A machine-readable oxygen-sweep table exists with oxygen bound, growth, ethanol flux, and yield.\", \"must_pass\": true}, {\"id\": \"req_carbon_source_table\", \"type\": \"artifact\", \"description\": \"A carbon-source comparison table exists, including unsupported sources marked explicitly.\", \"must_pass\": true}, {\"id\": \"req_figure\", \"type\": \"artifact\", \"description\": \"At least one oxygen-vs-ethanol or carbon-source yield figure exists with units.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"The ethanol-yield response to oxygen and carbon source is discussed mechanistically — why oxygen restriction increases ethanol yield and how carbon-source metabolism shapes fermentation. Acceptable inside summary, structured_results, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B04\", \"requirements\": \"A credible S. cerevisiae ethanol-yield study that loads a yeast GSMM, validates growth, sweeps oxygen uptake, swaps carbon sources, and reports ethanol secretion/yield with interpretable figures.\", \"judging_note\": \"The main challenge is adapting the pipeline beyond E. coli while remaining within COBRApy/BIGG capabilities. Accept iMM904 or another documented yeast GSMM.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b04-code\", \"requirements\": \"The code implements yeast model loading, medium setup, oxygen sweep, carbon-source swap, and ethanol-yield calculations.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b04-code-model\", \"requirements\": \"Loads iMM904 or comparable yeast model, identifies biomass, glucose, oxygen, and ethanol exchange reactions, and validates positive reference growth.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b04-code-o2\", \"requirements\": \"Sweeps oxygen uptake across at least 8 values while keeping substrate uptake defined and restoring state between runs.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b04-code-carbon\", \"requirements\": \"Compares at least 4 carbon sources where model reactions exist and explicitly records unsupported sources.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b04-code-yield\", \"requirements\": \"Calculates ethanol yield using absolute substrate uptake and separates yield from raw secretion flux.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b04-exec\", \"requirements\": \"Execution produces oxygen-sweep and carbon-source artifacts without crashing.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b04-exec-status\", \"requirements\": \"Reference glucose condition is optimal and at least 80% of supported oxygen/carbon conditions solve or are clearly marked infeasible.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b04-exec-physics\", \"requirements\": \"Mass-balance check passes for the yeast GSMM; no non-physical fluxes appear under aerobic glucose reference (e.g. ethanol secretion should be low under full aerobic conditions); biomass flux is positive under yeast glucose-minimal reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b04-exec-artifacts\", \"requirements\": \"Writes results.json, oxygen_sweep.csv, and carbon_source_comparison.csv or equivalent structured outputs.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b04-results\", \"requirements\": \"Results quantify oxygen and substrate effects on ethanol secretion/yield.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b04-result-h1\", \"requirements\": \"Reports yeast reference growth with model ID and medium definition. Score 100% if positive yeast biomass flux and ethanol secretion are both reported under aerobic glucose-minimal medium with optimal solver status and a named model ID; 67% if growth is positive but ethanol is not reported; 33% if feasibility is stated without numeric values; 0% otherwise.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b04-result-h2\", \"requirements\": \"Quantifies how ethanol flux or yield changes as oxygen decreases and identifies the low-oxygen fermentation regime. Score 100% if ethanol flux or yield is non-decreasing as oxygen uptake decreases across ≥6 of 8 oxygen-sweep points; 67% if non-decreasing over ≥4 of 8 points; 33% if the trend is described without a quantitative grid; 0% otherwise.\", \"weight\": 12, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b04-result-h3\", \"requirements\": \"Compares growth and ethanol yield across carbon sources with a biological interpretation. Score 100% if at least 3 carbon sources are compared with distinct numeric ethanol yields and a biological interpretation; 67% if 2 sources are compared with numeric differences; 33% if the comparison is qualitative or only one carbon source produces positive growth; 0% otherwise.\", \"weight\": 10, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b04-result-figure\", \"requirements\": \"Produces labelled figures for oxygen sweep and/or carbon-source yield comparison.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b04-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, names the oxygen regime and carbon source that maximise ethanol yield, and gives a one-paragraph mechanistic interpretation of why oxygen restriction switches yeast from respiratory growth to fermentative ethanol production.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b04-repro\", \"requirements\": \"Model source, medium, oxygen grid, carbon-source reactions, solver, and versions are documented.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b04-repro-model\", \"requirements\": \"The loaded yeast GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID, version, and source URL in the writeup; a downstream user can locate the exact model object used.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b04-repro-config\", \"requirements\": \"Saves reaction IDs, bounds, grid values, and carbon-source list in config or results metadata.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b04-repro-version\", \"requirements\": \"Records COBRApy, solver, model source/version, and rerun consistency for reference growth.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B04.yaml", "rubric_file": "tasks/biology/rubrics/B04.json"}
+{"id": "B05", "domain": "biology", "title": "M. tuberculosis condition-specific essentiality and metabolic drug-target prioritisation", "topic": "M. tuberculosis condition-specific essentiality and metabolic drug-target prioritisation", "domains": ["systems-biology", "drug-target-prioritisation", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 2400, "synthesis": "Constraint-based essentiality analysis can prioritise metabolic drug-target\nhypotheses in pathogen models. A credible study loads an M. tuberculosis GSMM\nsuch as iNJ661 when available, validates biomass production under a documented\nmedium, performs single-gene and/or single-reaction deletion, and ranks\nessential targets by growth impact and subsystem interpretability.", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"The selected M. tuberculosis model produces positive biomass under the documented reference medium.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"Single-gene or single-reaction deletion identifies a non-empty set of essential metabolic targets under the reference condition.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Prioritised targets are enriched in interpretable core metabolic subsystems such as cell-wall precursor, cofactor, lipid, energy, or amino-acid metabolism.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Which condition-specific metabolic genes or reactions are predicted essential in M. tuberculosis and are plausible drug-target hypotheses?\", \"conditions\": [{\"name\": \"mtb_reference_medium\", \"description\": \"M. tuberculosis iNJ661 or comparable model under a documented reference medium.\"}, {\"name\": \"single_gene_deletion\", \"description\": \"Single-gene deletion screen when GPR rules are available.\"}, {\"name\": \"single_reaction_deletion\", \"description\": \"Single-reaction deletion screen as a fallback or complementary target set.\"}], \"baselines\": [\"Wild-type reference growth.\", \"Nonessential deletion growth distribution.\"], \"metrics\": [{\"name\": \"wt_growth_rate_1_per_h\", \"direction\": \"maximize\", \"description\": \"Reference biomass flux.\"}, {\"name\": \"growth_fraction_after_deletion\", \"direction\": \"minimize\", \"description\": \"Deletion biomass divided by WT biomass.\"}, {\"name\": \"essential_target_count\", \"direction\": \"characterize\", \"description\": \"Count of genes/reactions below essentiality threshold.\"}, {\"name\": \"subsystem_enrichment\", \"direction\": \"characterize\", \"description\": \"Subsystem distribution among essential targets.\"}], \"datasets\": [{\"name\": \"M. tuberculosis GSMM\", \"source\": \"BIGG iNJ661 or documented local SBML/JSON model.\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 2400}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"results.json contains model ID, WT growth, essential target counts, top targets, and hypothesis verdicts.\", \"must_pass\": true}, {\"id\": \"req_deletion_table\", \"type\": \"artifact\", \"description\": \"A gene or reaction deletion table exists with target ID, status, growth, growth_fraction, and essential flag.\", \"must_pass\": true}, {\"id\": \"req_top_targets\", \"type\": \"discussion\", \"description\": \"Top 10 prioritised essential targets are reported with subsystem/pathway and rationale.\", \"must_pass\": true}, {\"id\": \"req_figure\", \"type\": \"artifact\", \"description\": \"At least one essentiality distribution or subsystem-enrichment figure exists.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"Top-3 essential target hypotheses are framed mechanistically — naming the metabolic subsystem each target belongs to and why its disruption is predicted lethal under the reference medium. Claims must be framed as computational hypotheses, not validated drugs. Acceptable inside summary, structured_results, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B05\", \"requirements\": \"A credible M. tuberculosis essentiality study using COBRApy that validates a pathogen GSMM, runs single-gene and/or single-reaction deletions, identifies essential targets, and prioritises interpretable drug-target hypotheses.\", \"judging_note\": \"Accept gene deletions, reaction deletions, or both depending on model GPR support. Penalize unsupported biological claims more than absence of external drug databases.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b05-code\", \"requirements\": \"The code loads the pathogen model, validates medium, runs deletion screens, and annotates essential targets.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b05-code-model\", \"requirements\": \"Loads iNJ661 or a documented M. tuberculosis GSMM, defines medium/objective, and validates positive WT biomass.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b05-code-deletion\", \"requirements\": \"Runs single-gene deletion when GPR rules are usable, or single-reaction deletion as an explicit fallback/complement.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b05-code-threshold\", \"requirements\": \"Defines essentiality threshold such as growth <5% or <10% of WT and applies it consistently.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b05-code-annotation\", \"requirements\": \"Extracts subsystem, reaction name, gene name, GPR, or other model-native annotations for top targets.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b05-exec\", \"requirements\": \"Deletion runs complete with interpretable status handling and structured outputs.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b05-exec-status\", \"requirements\": \"WT reference condition is optimal and at least 85% of deletion simulations solve or are explicitly marked infeasible/lethal.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b05-exec-physics\", \"requirements\": \"Mass-balance check passes for the loaded pathogen GSMM; WT biomass is positive under the documented medium; deletion table contains no negative growth values; model GPR rules produce logically consistent essential gene predictions. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b05-exec-artifacts\", \"requirements\": \"Writes results.json plus deletion table with growth, growth_fraction, status, and essential flag.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b05-results\", \"requirements\": \"Results identify essential targets and provide conservative biological interpretation.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b05-result-h1\", \"requirements\": \"Reports WT growth, medium, model ID, and validation caveats. Score 100% if positive WT biomass flux is reported under the documented medium with optimal solver status and model ID named; 67% if growth is positive but medium is not explicitly documented; 33% if feasibility is stated without a numeric value; 0% otherwise.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b05-result-h2\", \"requirements\": \"Reports essential target count and distribution of growth fractions after deletion. Score 100% if ≥5 essential genes/reactions are identified with growth fractions below the stated threshold and the deletion table covers ≥50 candidates; 67% if essential targets are identified but the candidate set covers fewer than 50 entries; 33% if essential targets are named without a deletion table; 0% otherwise.\", \"weight\": 10, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b05-result-h3\", \"requirements\": \"Top 10 targets include subsystem/pathway and one-sentence rationale; claims are framed as modelling hypotheses, not validated drugs. Score 100% if ≥3 of the top-10 prioritised targets belong to named core metabolic subsystems (cell-wall precursor, cofactor, lipid, energy, or amino-acid biosynthesis) with at least a one-sentence pathway rationale; 67% if 2 of 10 with rationale; 33% if 1 of 10 with rationale; 0% otherwise.\", \"weight\": 12, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b05-result-figure\", \"requirements\": \"Produces a labelled essentiality histogram, target ranking plot, or subsystem enrichment chart.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b05-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, names the top prioritised essential targets with subsystem/enzyme identity, and gives a one-paragraph mechanistic rationale for why each target's disruption is lethal. Claims are explicitly framed as computational hypotheses requiring wet-lab validation.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b05-repro\", \"requirements\": \"Model, medium, deletion type, threshold, solver, and annotations are reproducibly recorded.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b05-repro-model\", \"requirements\": \"The loaded M. tuberculosis GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID, version, or local file hash in the writeup; a downstream user can locate the exact model object used.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b05-repro-config\", \"requirements\": \"Saves model source, medium bounds, objective, deletion mode, and essentiality threshold.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b05-repro-version\", \"requirements\": \"Records COBRApy/solver versions and confirms rerun consistency for WT growth and essential count.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B05.yaml", "rubric_file": "tasks/biology/rubrics/B05.json"}
+{"id": "B06", "domain": "biology", "title": "E. coli carbon-source robustness and condition-dependent essential genes", "topic": "E. coli carbon-source robustness and condition-dependent essential genes", "domains": ["systems-biology", "essentiality-analysis", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 2400, "synthesis": "Carbon-source swaps are a direct mfa-agent capability and produce interpretable\ncondition-dependent phenotypes. A credible study loads an E. coli GSMM, defines\na base minimal medium, tests multiple single-carbon sources, runs essentiality\nanalysis on a focused gene set under each feasible source, and identifies genes\nwhose essentiality changes with carbon source.", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"The model predicts positive growth on glucose and at least two additional supported carbon sources.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"Growth rates and secretion profiles differ substantially across carbon sources.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"A focused central-carbon gene set contains condition-dependent essential genes whose deletion effects vary by carbon source.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Which E. coli metabolic vulnerabilities are carbon-source dependent across glucose, glycerol, acetate, succinate, and fructose-like conditions?\", \"conditions\": [{\"name\": \"carbon_source_panel\", \"description\": \"Close background carbon uptake, open one carbon source at a time, and run FBA/pFBA.\"}, {\"name\": \"focused_gene_essentiality_by_carbon\", \"description\": \"For feasible carbon sources, delete a focused set of central-carbon genes and record growth fractions.\"}], \"baselines\": [\"Glucose-minimal medium.\", \"Wild-type growth for each carbon source.\"], \"metrics\": [{\"name\": \"growth_rate_by_carbon\", \"direction\": \"characterize\", \"description\": \"WT biomass flux per carbon source.\"}, {\"name\": \"major_secretions_by_carbon\", \"direction\": \"characterize\", \"description\": \"Positive exchange fluxes under pFBA.\"}, {\"name\": \"condition_dependent_essential_count\", \"direction\": \"maximize\", \"description\": \"Genes essential under one source but not another.\"}], \"datasets\": [{\"name\": \"E. coli GSMM\", \"source\": \"BIGG iJO1366 or iAF1260 via cobra.io.load_model\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 2400}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"results.json contains carbon-source growth table summary, essentiality counts, hypothesis verdicts, and summary.\", \"must_pass\": true}, {\"id\": \"req_carbon_table\", \"type\": \"artifact\", \"description\": \"A carbon-source comparison table exists with source, exchange reaction, status, growth, and secretion profile fields.\", \"must_pass\": true}, {\"id\": \"req_essentiality_matrix\", \"type\": \"artifact\", \"description\": \"A gene-by-carbon essentiality/growth-fraction matrix exists for feasible carbon sources.\", \"must_pass\": true}, {\"id\": \"req_heatmap\", \"type\": \"artifact\", \"description\": \"A heatmap or clustered plot of condition-dependent gene essentiality exists with labelled axes.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"Top-3 condition-dependent essential genes are explained mechanistically — why each gene is essential under one carbon source but not another, referencing the metabolic pathway context. Acceptable inside summary, structured_results, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B06\", \"requirements\": \"A credible E. coli carbon-source robustness study that loads a validated GSMM, swaps carbon sources, reports growth/secretion profiles, and identifies condition-dependent essential genes from a focused gene set.\", \"judging_note\": \"Score explicit medium handling and condition-dependent interpretation. It is acceptable if some carbon sources are unsupported, provided they are recorded and not silently dropped.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b06-code\", \"requirements\": \"The code implements carbon-source medium swaps, FBA/pFBA, and focused essentiality comparison.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b06-code-model\", \"requirements\": \"Loads E. coli BIGG model, identifies biomass and candidate carbon exchange reactions, and defines a minimal medium policy.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b06-code-swap\", \"requirements\": \"Closes background carbon uptake and opens one carbon source at a time with documented uptake bounds.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"b06-code-secretion\", \"requirements\": \"Runs FBA/pFBA per source and records major positive exchange fluxes.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b06-code-essentiality\", \"requirements\": \"Runs focused single-gene deletions for feasible carbon sources and computes growth fractions relative to each source-specific WT.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b06-exec\", \"requirements\": \"Execution produces carbon-source and essentiality artifacts.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b06-exec-validity\", \"requirements\": \"Glucose reference is optimal and at least two non-glucose sources are either feasible or clearly marked unsupported/infeasible.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b06-exec-physics\", \"requirements\": \"Mass-balance check passes for the E. coli GSMM under each tested carbon source; no carbon source produces negative biomass or non-physical secretion without explicit handling; model state is restored between carbon-source conditions. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b06-exec-artifacts\", \"requirements\": \"Writes results.json, carbon-source table, and gene-by-condition growth-fraction matrix.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b06-results\", \"requirements\": \"Results compare carbon sources and identify condition-dependent vulnerabilities.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b06-result-h1\", \"requirements\": \"Reports feasible carbon-source count and growth rates, including unsupported-source handling. Score 100% if ≥3 carbon sources including glucose support positive biomass growth with explicit exchange bounds documented; 67% if 2 sources support positive growth; 33% if only glucose is feasible with explicit confirmation that other sources are infeasible or unsupported; 0% otherwise.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b06-result-h2\", \"requirements\": \"Compares secretion profiles and growth rates across sources with quantitative differences. Score 100% if growth rates differ by >10% between at least 2 feasible carbon sources AND major secretion fluxes differ qualitatively across sources; 67% if growth rate differences are reported without secretion-profile comparison; 33% if differences are noted qualitatively without quantification; 0% otherwise.\", \"weight\": 9, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b06-result-h3\", \"requirements\": \"Identifies genes with large growth-fraction changes across carbon sources and explains pathway context for top hits. Score 100% if ≥3 genes with condition-dependent essentiality changes (essential under one carbon source but not another) are identified with pathway context; 67% if 2 genes are identified; 33% if 1 gene is identified with pathway context; 0% otherwise.\", \"weight\": 12, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b06-result-figure\", \"requirements\": \"Produces labelled bar plots and/or heatmaps for growth, secretion, and condition-dependent essentiality.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b06-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, names supported and unsupported carbon sources, identifies condition-dependent essential genes, and gives a one-paragraph mechanistic interpretation linking carbon-source specific metabolism to the observed vulnerability changes.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b06-repro\", \"requirements\": \"Carbon-source definitions, gene set, thresholds, solver, and model are documented.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b06-repro-model\", \"requirements\": \"The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce each carbon-source medium setup.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b06-repro-runcard\", \"requirements\": \"Saves carbon exchange IDs, uptake bounds, oxygen setting, gene list, and essentiality threshold.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b06-repro-version\", \"requirements\": \"Records COBRApy/solver versions and rerun consistency for glucose and at least one non-glucose source.\", \"weight\": 5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B06.yaml", "rubric_file": "tasks/biology/rubrics/B06.json"}
+{"id": "B07", "domain": "biology", "title": "Reproducible FBA protocol benchmark: FBA vs pFBA vs loopless/FVA on E. coli", "topic": "Reproducible FBA protocol benchmark: FBA vs pFBA vs loopless/FVA on E. coli", "domains": ["systems-biology", "method-benchmarking", "constraint-based-modelling"], "arxiv_id": null, "venue": "ARC-Bench Biology 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 1800, "synthesis": "A method benchmark is well matched to mfa-agent because it tests robust use of\nCOBRApy APIs rather than biological novelty alone. A credible study loads a\nvalidated E. coli GSMM, runs standard FBA, pFBA, loopless FBA when available,\nand FVA at a fixed fraction of optimum, then compares growth, flux sparsity,\nruntime, and reaction-level variability.", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"FBA, pFBA, and loopless FBA preserve the same biomass optimum within solver tolerance under the same medium.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"pFBA produces a lower total absolute flux norm and/or fewer active reactions than unconstrained FBA.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"FVA reveals that only a subset of central-carbon reactions are tightly constrained near optimum while others remain variable.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"How do standard FBA, pFBA, loopless FBA, and FVA differ in flux parsimony and variability while preserving E. coli growth predictions?\", \"conditions\": [{\"name\": \"standard_fba\", \"description\": \"Maximize biomass on aerobic glucose-minimal medium.\"}, {\"name\": \"pfba\", \"description\": \"Maximize biomass then minimize total absolute flux.\"}, {\"name\": \"loopless_fba\", \"description\": \"Run loopless solution when available and compare objective/fluxes.\"}, {\"name\": \"fva_95_percent_growth\", \"description\": \"Run FVA at 95% of optimum and classify rigid vs flexible reactions.\"}], \"baselines\": [\"Standard FBA biomass optimum.\", \"COBRApy solver status and runtime.\"], \"metrics\": [{\"name\": \"growth_rate_1_per_h\", \"direction\": \"match\", \"description\": \"Biomass flux for each method.\"}, {\"name\": \"total_abs_flux\", \"direction\": \"minimize\", \"description\": \"Sum of absolute reaction fluxes.\"}, {\"name\": \"active_reaction_count\", \"direction\": \"minimize\", \"description\": \"Number of reactions with absolute flux above tolerance.\"}, {\"name\": \"fva_width\", \"direction\": \"characterize\", \"description\": \"Maximum minus minimum FVA interval per reaction.\"}], \"datasets\": [{\"name\": \"E. coli GSMM\", \"source\": \"BIGG iJO1366 or iAF1260 via cobra.io.load_model\"}], \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 1800}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"results.json contains method-level metrics, hypotheses, summary, and solver/version metadata.\", \"must_pass\": true}, {\"id\": \"req_method_table\", \"type\": \"artifact\", \"description\": \"A method-comparison table exists with method, status, growth, total_abs_flux, active_reactions, and runtime.\", \"must_pass\": true}, {\"id\": \"req_fva_table\", \"type\": \"artifact\", \"description\": \"An FVA table exists with reaction ID, minimum, maximum, width, and rigid/flexible classification.\", \"must_pass\": true}, {\"id\": \"req_figures\", \"type\": \"artifact\", \"description\": \"At least one labelled figure compares methods or FVA variability.\", \"must_pass\": true}, {\"id\": \"req_h1_h2_h3_supported_flags\", \"type\": \"discussion\", \"description\": \"Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST have an explicit `supported` boolean and a `details` string of at least 40 characters describing the evidence used to reach that verdict.\\n\", \"must_pass\": true}, {\"id\": \"req_mechanistic_writeup\", \"type\": \"discussion\", \"description\": \"The protocol comparison is summarised with a method recommendation — which method is most reproducible and why, referencing flux-norm and FVA evidence. Acceptable inside summary, method_comparison, or a separate writeup field.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"B07\", \"requirements\": \"A reproducible FBA-method benchmark on E. coli that compares standard FBA, pFBA, loopless FBA when available, and FVA near optimum using consistent medium, metrics, runtime logging, and interpretable figures.\", \"judging_note\": \"This is a protocol-quality task. Reward robust implementation, structured outputs, and careful handling of unavailable loopless solvers over biological novelty.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"b07-code\", \"requirements\": \"The code implements method comparison with consistent model state and metrics.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b07-code-model\", \"requirements\": \"Loads E. coli BIGG model, sets medium and biomass objective, and validates reference FBA.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b07-code-methods\", \"requirements\": \"Runs standard FBA, pFBA, and loopless FBA or records a justified loopless-unavailable status.\", \"weight\": 7, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b07-code-fva\", \"requirements\": \"Runs FVA at a documented fraction of optimum and computes interval widths and rigid/flexible labels.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"b07-code-metrics\", \"requirements\": \"Computes total absolute flux, active reaction count, objective value, solver status, and runtime for each method.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b07-exec\", \"requirements\": \"Execution produces complete benchmark artifacts and handles optional method failures explicitly.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b07-exec-status\", \"requirements\": \"Standard FBA and pFBA solve optimally; loopless and FVA either solve or have explicit, non-crashing failure records.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b07-exec-physics\", \"requirements\": \"Mass-balance check passes for the loaded model under the benchmark medium; FBA, pFBA, and loopless FBA all return non-negative biomass fluxes; no negative flux norms or ill-conditioned solver status appear without explicit logging. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.\", \"weight\": 7.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b07-exec-artifacts\", \"requirements\": \"Writes results.json, method comparison table, FVA table, and method flux tables.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b07-results\", \"requirements\": \"Results compare objective preservation, parsimony, and reaction variability.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"b07-result-h1\", \"requirements\": \"Reports objective differences among methods and checks agreement within solver tolerance where methods succeed. Score 100% if FBA and pFBA biomass objectives agree within 1% relative error and any available loopless FBA also agrees within 1%; 67% if FBA and pFBA agree but loopless comparison is missing or failed with explicit explanation; 33% if only qualitative agreement is stated; 0% otherwise.\", \"weight\": 9, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"b07-result-h2\", \"requirements\": \"Quantifies pFBA parsimony using total flux norm and active reaction count compared with standard FBA. Score 100% if pFBA total absolute flux norm is numerically lower than FBA AND active reaction count is lower under pFBA; 67% if flux norm is lower but active reaction count is not compared; 33% if pFBA parsimony is stated qualitatively without numeric comparison; 0% otherwise.\", \"weight\": 10, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b07-result-h3\", \"requirements\": \"Summarizes FVA width distribution and names central-carbon reactions that are rigid or flexible near optimum. Score 100% if FVA interval widths are reported for all model reactions and ≥5 central-carbon reactions are classified as rigid (interval width < 1 mmol/gDW/h) with subsystem identification; 67% if rigid/flexible classification is reported without central-carbon specifics; 33% if FVA results are reported without classification; 0% otherwise.\", \"weight\": 10, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b07-result-figure\", \"requirements\": \"Produces labelled method-comparison and/or FVA-width figures with units and clear legends.\", \"weight\": 8, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"b07-result-writeup\", \"requirements\": \"The README or results writeup discusses each hypothesis outcome, summarises the method-level comparison (parsimony, variability, runtime), and gives a one-paragraph protocol recommendation stating which method is most suitable for reproducible phenotype prediction and why.\", \"weight\": 6.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"b07-repro\", \"requirements\": \"The benchmark is reproducible with pinned model, medium, tolerance, solver, and code path.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"b07-repro-model\", \"requirements\": \"The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce the benchmark medium.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b07-repro-config\", \"requirements\": \"Saves model ID, objective, medium bounds, FVA fraction, active-flux tolerance, and method list.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}, {\"id\": \"b07-repro-version\", \"requirements\": \"Records COBRApy, optlang, solver backend/version, and runtime environment.\", \"weight\": 6, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Reproducibility\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/biology/manifests/B07.yaml", "rubric_file": "tasks/biology/rubrics/B07.json"}

data/statistics.jsonl

@@ -0,0 +1,3 @@
+{"id": "S01", "domain": "statistics", "title": "Bootstrap confidence intervals under non-Gaussian data", "topic": "Bootstrap confidence interval coverage under light-tailed, skewed, and heavy-tailed data", "domains": ["statistics", "resampling-methods", "simulation-study"], "arxiv_id": null, "venue": "ARC-Bench Statistics 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 300, "synthesis": "Bootstrap confidence intervals are widely used when analytic standard errors\nare inconvenient, but their finite-sample coverage can change substantially\nwith skewness, heavy tails, contamination, and the choice of statistic. A\ncredible study of this topic should simulate data-generating processes with\nknown estimands, compare several bootstrap confidence interval methods, and\nevaluate empirical coverage and interval width.\n\nThe study should not only report whether nominal 95% intervals contain the\ntruth, but also explain when wider intervals, robust estimators, or\nstudentized intervals improve reliability. The research question is: *how do\nbootstrap interval choices and estimator choices affect empirical coverage\nunder light-tailed, skewed, and heavy-tailed data?*", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"Percentile bootstrap intervals for the sample mean achieve near-nominal coverage under light-tailed Gaussian data, especially at moderate sample sizes.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"Percentile bootstrap intervals for the sample mean under-cover under heavy-tailed or contaminated data.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Robust location estimators, such as the median or trimmed mean, improve coverage stability or interval behavior under heavy-tailed or contaminated data.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"How do bootstrap confidence interval methods behave across light-tailed, skewed, and heavy-tailed distributions?\", \"estimands\": [{\"name\": \"population_mean\", \"description\": \"The true mean of each simulated data-generating process when it exists.\"}, {\"name\": \"robust_location\", \"description\": \"A robust target such as the median or trimmed mean, used when heavy tails or contamination make mean-based inference unstable.\"}], \"conditions\": [{\"name\": \"normal_theory_mean_ci\", \"description\": \"Classical normal or t-based interval for the sample mean.\"}, {\"name\": \"percentile_bootstrap_mean\", \"description\": \"Percentile bootstrap interval for the sample mean.\"}, {\"name\": \"studentized_bootstrap_mean\", \"description\": \"Studentized or bootstrap-t interval for the sample mean, or a justified bootstrap alternative.\"}, {\"name\": \"robust_estimator_bootstrap\", \"description\": \"Bootstrap interval for a robust estimator such as the median or trimmed mean.\"}], \"data_generating_conditions\": [{\"name\": \"gaussian\", \"description\": \"Light-tailed data where standard mean inference should work well.\"}, {\"name\": \"lognormal_skewed\", \"description\": \"Skewed data where percentile intervals may show asymmetric finite-sample behavior.\"}, {\"name\": \"student_t_heavy_tailed\", \"description\": \"Heavy-tailed data with unstable sample means.\"}, {\"name\": \"contaminated_normal\", \"description\": \"Mostly normal data with a small fraction of extreme outliers.\"}], \"metrics\": [{\"name\": \"coverage\", \"direction\": \"target_0.95\", \"description\": \"Empirical fraction of intervals containing the true estimand.\"}, {\"name\": \"coverage_error\", \"direction\": \"minimize_abs\", \"description\": \"Absolute deviation from nominal 95% coverage.\"}, {\"name\": \"interval_width\", \"direction\": \"minimize_given_coverage\", \"description\": \"Average interval width, interpreted jointly with coverage.\"}, {\"name\": \"failure_rate\", \"direction\": \"minimize\", \"description\": \"Fraction of simulation runs where an interval could not be computed or produced invalid values.\"}], \"simulation_settings\": {\"sample_sizes\": [30, 100, 500], \"monte_carlo_repetitions\": 500, \"bootstrap_resamples\": 1000, \"seeds\": [0, 1, 2, 3, 4]}, \"expected_artifacts\": [{\"path\": \"src/experiment.py\", \"description\": \"Executable simulation code implementing confidence interval methods and data-generating processes.\"}, {\"path\": \"results/metrics.json\", \"description\": \"Machine-readable coverage, coverage error, interval width, and failure rate by method, distribution, and sample size.\"}, {\"path\": \"results/figures/\", \"description\": \"Diagnostic plots comparing coverage and interval width across methods and data-generating conditions.\"}, {\"path\": \"report/paper.md\", \"description\": \"Paper-style report explaining the setup, methods, results, limitations, and per-hypothesis conclusions.\"}], \"dependencies\": {\"python\": [\"numpy\", \"pandas\", \"scipy\", \"matplotlib\"]}, \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 300}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"A canonical results.json file exists at the workspace root with at least the keys: primary_metric (number), metric_key (string), metrics (object with numeric keys), hypotheses (object with h1/h2/h3 entries each carrying a `supported` boolean), summary (non-empty string).\", \"must_pass\": true}, {\"id\": \"req_coverage_metrics_table\", \"type\": \"artifact\", \"description\": \"A machine-readable metrics artifact (e.g. results/metrics.json) exists reporting empirical coverage AND interval width broken down by CI method, data-generating distribution, and sample size.\", \"must_pass\": true}, {\"id\": \"req_coverage_numeric\", \"type\": \"numeric\", \"description\": \"results.json metrics MUST contain at least 3 numeric (non-null, finite) empirical coverage values in [0, 1] for distinct (method, distribution) cells, including at least one Gaussian cell and one heavy-tailed or contaminated cell so H1 and H2 can be evaluated.\", \"must_pass\": true}, {\"id\": \"req_hypotheses_supported_flags\", \"type\": \"discussion\", \"description\": \"results.json hypotheses.h1/h2/h3 each MUST have an explicit `supported` boolean AND a `details` string ≥ 40 characters quoting the numerical coverage / interval-width evidence (specific values + their source artifact) used to reach the verdict.\", \"must_pass\": true}, {\"id\": \"req_diagnostic_figure\", \"type\": \"artifact\", \"description\": \"At least one figure file (PNG or PDF, ≥150 DPI for raster) exists under results/figures/ comparing coverage and/or interval width across methods and data-generating conditions, with axes labeled and a legend.\", \"must_pass\": false}, {\"id\": \"req_width_tradeoff_writeup\", \"type\": \"discussion\", \"description\": \"The report discusses the coverage-versus-interval-width tradeoff — recognising that higher coverage may simply come from wider intervals and that width must be read jointly with coverage. Nice-to-have, not blocking.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"results.json or a sibling reproducibility section names the Monte Carlo repetition count, bootstrap resample count, and at least one explicit random seed. Required for full reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"s01-root\", \"requirements\": \"A credible simulation study evaluating whether bootstrap confidence interval choice affects empirical coverage under light-tailed, skewed, and heavy-tailed data. The submission should implement multiple CI methods, simulate several data-generating distributions and sample sizes, report coverage and interval width over repeated trials, and connect the numeric findings to the three hypotheses.\", \"judging_note\": \"Score on statistical substance, correct simulation design, and directional correctness of evidence. Do not require exact repetition counts if the submission is computationally reasonable. Partial but well-motivated simulations deserve partial credit; rigid naming differences should not penalize a substantively correct experiment.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"s01-code\", \"requirements\": \"The bootstrap simulation code implements meaningful confidence interval comparisons across relevant distributional regimes.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s01-code-ci-methods\", \"requirements\": \"The submission implements at least two bootstrap confidence interval methods, typically percentile bootstrap and studentized/bootstrap-t or another justified alternative. Methods should be implemented as distinct code paths rather than cosmetic options.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s01-code-robust-estimator\", \"requirements\": \"The submission includes a robust location estimator, such as median or trimmed mean, and compares it against the ordinary sample mean under heavy-tailed or contaminated data.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s01-code-dgps\", \"requirements\": \"The submission simulates multiple data-generating processes covering at least light-tailed and heavy-tailed cases, with skewed data such as lognormal or contaminated data receiving additional credit.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Dataset and Model Acquisition\"}, {\"id\": \"s01-code-sample-sizes\", \"requirements\": \"The simulation evaluates more than one sample size so that finite-sample behavior can be distinguished from asymptotic behavior.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s01-exec\", \"requirements\": \"Execution produces coverage and interval-quality metrics adequate to evaluate bootstrap CI behavior.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s01-exec-metrics\", \"requirements\": \"Execution produces a machine-readable metrics artifact, such as results/metrics.json, containing numeric empirical coverage and interval width by condition, distribution, and sample size. Coverage error or failure rate may supplement these metrics.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s01-exec-repetitions\", \"requirements\": \"Reported metrics are based on repeated Monte Carlo trials and bootstrap resampling. The exact number of repetitions need not match the topic file, but it should be large enough to make coverage comparisons meaningful and should be honestly reported.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s01-exec-truth\", \"requirements\": \"The simulation correctly defines the true estimand for each data-generating process, such as the true mean or true robust location target, and checks whether confidence intervals contain that estimand.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s01-paper\", \"requirements\": \"The final paper or report addresses the three bootstrap hypotheses with quantitative evidence and a clear statistical narrative.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"s01-result-h1\", \"requirements\": \"The submission evaluates whether percentile bootstrap confidence intervals achieve near-nominal coverage under light-tailed Gaussian data, especially at moderate or larger sample sizes, and states whether H1 is supported, refuted, or inconclusive.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s01-result-h2\", \"requirements\": \"The submission evaluates whether percentile bootstrap intervals for the sample mean undercover under heavy-tailed data and supports the conclusion with empirical coverage numbers.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s01-result-h3\", \"requirements\": \"The submission compares robust location estimators against the ordinary sample mean under heavy-tailed or contaminated data and discusses whether robustness improves coverage stability or interval behavior.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s01-result-width-tradeoff\", \"requirements\": \"The analysis discusses interval-width tradeoffs, recognizing that higher coverage may come from wider intervals and that width should be interpreted jointly with coverage.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s01-result-writeup\", \"requirements\": \"The README or writeup describes the simulation setup, CI methods, data-generating processes, key numeric results, and per-hypothesis outcomes with appropriate caveats on repetition count, bootstrap count, and Monte Carlo uncertainty.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 12, "manifest_file": "tasks/statistics/manifests/S01.yaml", "rubric_file": "tasks/statistics/rubrics/S01.json"}
+{"id": "S02", "domain": "statistics", "title": "Double machine learning for average treatment effect estimation", "topic": "Double machine learning for average treatment effect estimation under nonlinear confounding", "domains": ["statistics", "causal-inference", "semiparametric-estimation"], "arxiv_id": null, "venue": "ARC-Bench Statistics 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 300, "synthesis": "Semiparametric causal inference separates a low-dimensional target parameter,\nsuch as the average treatment effect, from infinite-dimensional nuisance\nfunctions such as the propensity score e(X) and outcome regressions m0(X) and\nm1(X). Double machine learning studies how the target parameter can remain\nestimable when those nuisance functions are learned flexibly rather than\nspecified by a finite-dimensional parametric model.\n\nThe benchmark should make this relationship explicit. The study should\nimplement and evaluate treatment effect estimators in simulated observational\ndata with nonlinear confounding, compare naive regression or inverse-propensity\nweighting against orthogonalized / doubly robust estimators, and test how\nNeyman orthogonality and cross-fitting reduce first-order sensitivity to\nnuisance estimation error. The research question is: *when the nuisance\nfunctions are treated as infinite-dimensional objects and estimated with\nflexible learners, does orthogonalization preserve reliable inference for the\nfinite-dimensional ATE?*", "num_hypotheses": 4, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"A naive difference-in-means estimator is biased under confounded treatment assignment.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"A doubly robust / orthogonalized estimator has lower absolute bias than simple plug-in regression or IPW when nuisance functions are nonlinear.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Cross-fitting improves stability or reduces bias compared with fitting nuisance functions and estimating the treatment effect on the same sample.\", \"measurable\": true}, {\"id\": \"H4\", \"statement\": \"Neyman-orthogonal scores make the ATE estimate less sensitive to nuisance-function estimation error than non-orthogonal plug-in estimators.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"How does double machine learning estimate a finite-dimensional ATE while using flexible estimators for infinite-dimensional nuisance functions under nonlinear confounding?\", \"estimand\": {\"name\": \"average_treatment_effect\", \"description\": \"ATE = E[Y(1) - Y(0)] in a simulated observational study with known ground truth.\"}, \"semiparametric_structure\": {\"target_parameter\": {\"name\": \"finite_dimensional_ate\", \"description\": \"The scalar ATE is the parameter of scientific interest.\"}, \"nuisance_functions\": [{\"name\": \"propensity_score\", \"notation\": \"e(X) = P(A = 1 | X)\", \"role\": \"Controls confounded treatment assignment and appears in IPW and AIPW scores.\"}, {\"name\": \"outcome_regression_treated\", \"notation\": \"m1(X) = E[Y | A = 1, X]\", \"role\": \"Models the treated potential-outcome regression.\"}, {\"name\": \"outcome_regression_control\", \"notation\": \"m0(X) = E[Y | A = 0, X]\", \"role\": \"Models the control potential-outcome regression.\"}], \"key_relationship\": \"The benchmark should evaluate whether an orthogonal score protects the finite-dimensional ATE estimate from first-order errors in flexible, effectively infinite-dimensional nuisance estimates.\"}, \"conditions\": [{\"name\": \"difference_in_means\", \"description\": \"Naive treated-control mean difference without confounding adjustment.\"}, {\"name\": \"ols_adjustment\", \"description\": \"Linear regression adjustment for covariates.\"}, {\"name\": \"ipw_logistic\", \"description\": \"Inverse propensity weighting using logistic regression propensity scores.\"}, {\"name\": \"aipw_no_crossfit\", \"description\": \"Augmented inverse propensity weighting using nuisance models fit on the same sample.\"}, {\"name\": \"dml_aipw_crossfit\", \"description\": \"Doubly robust AIPW estimator with K-fold cross-fitting and flexible nuisance models.\"}, {\"name\": \"nuisance_quality_sensitivity\", \"description\": \"A diagnostic condition comparing weaker and stronger nuisance learners to test sensitivity of each estimator to nuisance estimation error.\"}], \"nuisance_models\": [{\"name\": \"propensity_model\", \"candidates\": [\"logistic_regression\", \"random_forest_classifier\", \"gradient_boosting_classifier\"]}, {\"name\": \"outcome_model\", \"candidates\": [\"linear_regression\", \"random_forest_regressor\", \"gradient_boosting_regressor\"]}], \"metrics\": [{\"name\": \"bias\", \"direction\": \"minimize_abs\", \"description\": \"Mean estimated ATE minus true ATE across simulation repetitions.\"}, {\"name\": \"rmse\", \"direction\": \"minimize\", \"description\": \"Root mean squared error of ATE estimates.\"}, {\"name\": \"coverage\", \"direction\": \"target_0.95\", \"description\": \"Empirical coverage of nominal 95% confidence intervals.\"}, {\"name\": \"interval_width\", \"direction\": \"minimize_given_coverage\", \"description\": \"Average confidence interval width.\"}, {\"name\": \"estimate_std\", \"direction\": \"diagnostic\", \"description\": \"Monte Carlo standard deviation of ATE estimates.\"}], \"simulation_settings\": {\"sample_sizes\": [500, 1000, 3000], \"monte_carlo_repetitions\": 500, \"crossfit_folds\": 2, \"seeds\": [0, 1, 2, 3, 4]}, \"data_generating_process\": {\"covariates\": \"X ~ multivariate normal or uniform with nonlinear transformations\", \"treatment\": \"A ~ Bernoulli(sigmoid nonlinear function of X)\", \"outcome\": \"Y = tau * A + nonlinear function of X + noise\", \"true_ate\": 1.0}, \"expected_artifacts\": [{\"path\": \"src/experiment.py\", \"description\": \"Executable simulation code implementing all ATE estimators.\"}, {\"path\": \"results/metrics.json\", \"description\": \"Machine-readable bias, RMSE, coverage, and interval width by estimator and sample size.\"}, {\"path\": \"results/figures/\", \"description\": \"Plots or tables summarizing estimator bias, RMSE, coverage, interval width, and cross-fitting effects.\"}, {\"path\": \"report/paper.md\", \"description\": \"Paper-style report explaining the causal estimand, DGP, estimators, uncertainty method, results, limitations, and per-hypothesis conclusions.\"}, {\"path\": \"README.md\", \"description\": \"Setup and run instructions for reproducing the full experiment from a clean checkout.\"}], \"dependencies\": {\"python\": [\"numpy\", \"pandas\", \"scipy\", \"scikit-learn\", \"matplotlib\"]}, \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 300}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"A canonical results.json file exists at the workspace root with at least the keys: primary_metric (number), metric_key (string), metrics (object with numeric keys), hypotheses (object with h1/h2/h3/h4 entries each carrying a `supported` boolean), summary (non-empty string).\", \"must_pass\": true}, {\"id\": \"req_ate_metrics_table\", \"type\": \"artifact\", \"description\": \"A machine-readable metrics artifact (e.g. results/metrics.json) exists reporting ATE estimate, bias (or absolute bias), and RMSE broken down by estimator and sample size, with confidence-interval coverage or width where applicable.\", \"must_pass\": true}, {\"id\": \"req_bias_numeric\", \"type\": \"numeric\", \"description\": \"results.json metrics MUST contain numeric (non-null, finite) bias or absolute-bias values for at least the naive difference-in-means estimator and the doubly robust / cross-fitted DML estimator, evaluated against the known true ATE, so H1 and H2 can be evaluated quantitatively.\", \"must_pass\": true}, {\"id\": \"req_hypotheses_supported_flags\", \"type\": \"discussion\", \"description\": \"results.json hypotheses.h1/h2/h3/h4 each MUST have an explicit `supported` boolean AND a `details` string ≥ 40 characters quoting the numerical evidence (specific bias / RMSE / coverage values + their source artifact) used to reach the verdict.\", \"must_pass\": true}, {\"id\": \"req_crossfit_implemented\", \"type\": \"artifact\", \"description\": \"The DML condition implements sample splitting or K-fold cross-fitting so nuisance models are trained out-of-fold relative to the observations used in the orthogonal score; a non-cross-fitted AIPW comparison is also present so H3 can be evaluated.\", \"must_pass\": true}, {\"id\": \"req_estimand_writeup\", \"type\": \"discussion\", \"description\": \"The report clearly distinguishes the finite-dimensional causal estimand, the infinite-dimensional nuisance functions, the orthogonal score, and the evaluation metrics — avoiding confusion between prediction performance and ATE estimation quality. Nice-to-have, not blocking.\", \"must_pass\": false}, {\"id\": \"req_seed_documented\", \"type\": \"discussion\", \"description\": \"results.json or a sibling reproducibility section names the Monte Carlo repetition count, cross-fitting fold count, and at least one explicit random seed. Required for full reproducibility but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"s02-root\", \"requirements\": \"A credible semiparametric simulation study evaluating how double machine learning / orthogonalized AIPW estimates a finite-dimensional average treatment effect while using flexible estimators for infinite-dimensional nuisance functions under nonlinear confounding. The submission should implement multiple estimators, simulate observational data with known true ATE, evaluate bias, RMSE, confidence interval behavior, and sensitivity to nuisance estimation quality, and tie the numeric findings to the hypotheses.\", \"judging_note\": \"Score on causal and semiparametric substance rather than exact package choices. A correct hand-written AIPW / DML implementation should receive full credit even if it does not use a specialized causal inference library. Partial credit should reward clear separation of the finite-dimensional estimand from infinite-dimensional nuisance functions, correct use of orthogonal scores, nuisance estimation, cross-fitting, and honest uncertainty reporting.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"s02-code\", \"requirements\": \"The code implements a meaningful comparison of ATE estimators under simulated confounding with known ground truth.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s02-code-dgp\", \"requirements\": \"The submission implements a simulated observational data-generating process with covariates, confounded treatment assignment, outcome generation, known true average treatment effect, and explicit true or approximate nuisance functions such as propensity scores and outcome regressions.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Dataset and Model Acquisition\"}, {\"id\": \"s02-code-baselines\", \"requirements\": \"The submission implements simple baseline estimators such as difference-in-means, regression adjustment, and/or inverse propensity weighting so that DML is compared against meaningful alternatives.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s02-code-aipw\", \"requirements\": \"The submission implements a doubly robust or Neyman-orthogonal AIPW-style estimator using estimated propensity and outcome nuisance functions, and explains why the score targets the finite-dimensional ATE while treating those nuisances as flexible or infinite-dimensional.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s02-code-crossfit\", \"requirements\": \"The DML condition uses sample splitting or K-fold cross-fitting so nuisance models are trained out-of-fold relative to the observations used in the orthogonal score.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s02-code-nuisance-models\", \"requirements\": \"The nuisance functions are estimated with reasonable models for the simulated setting, such as logistic regression, random forests, gradient boosting, or other justified supervised learning methods, with at least one comparison that changes nuisance-model flexibility or quality.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s02-exec\", \"requirements\": \"Execution produces ATE estimation metrics adequate to evaluate bias, precision, and interval behavior.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s02-exec-metrics\", \"requirements\": \"Execution produces a machine-readable metrics artifact, such as results/metrics.json, containing numeric ATE estimates, bias or absolute bias, RMSE, and preferably confidence interval coverage or interval width by estimator and sample size.\", \"weight\": 17.5, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s02-exec-repetitions\", \"requirements\": \"Reported metrics are aggregated over multiple simulation repetitions or random seeds, with some dispersion reporting such as standard deviation, standard error, confidence interval, or quantiles.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s02-exec-sample-sizes\", \"requirements\": \"The experiment evaluates at least one nontrivial sample size and receives more credit for multiple sample sizes showing finite-sample behavior.\", \"weight\": 5.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s02-exec-uncertainty\", \"requirements\": \"The submission computes uncertainty estimates, such as influence-function standard errors, bootstrap standard errors, or Monte Carlo confidence intervals, sufficient to discuss coverage or reliability.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s02-paper\", \"requirements\": \"The final paper or report addresses the semiparametric hypotheses with quantitative evidence and a clear causal-inference narrative about finite-dimensional targets and infinite-dimensional nuisance functions.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"s02-result-h1\", \"requirements\": \"The submission evaluates whether the naive difference-in-means estimator is biased under confounded treatment assignment and supports the conclusion using the known true ATE.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s02-result-h2\", \"requirements\": \"The submission compares the doubly robust / orthogonalized estimator against simpler plug-in regression or IPW estimators and states whether DML reduces absolute bias or RMSE under nonlinear nuisance functions.\", \"weight\": 17.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s02-result-h3\", \"requirements\": \"The submission compares cross-fitted and non-cross-fitted versions of the doubly robust estimator, or otherwise clearly analyzes the contribution of cross-fitting to bias, RMSE, or stability.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s02-result-h4\", \"requirements\": \"The submission evaluates whether Neyman-orthogonal scores make the ATE estimate less sensitive to nuisance-function estimation error than non-orthogonal plug-in estimators, using weaker versus stronger nuisance learners or another explicit nuisance-quality diagnostic.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s02-result-estimand\", \"requirements\": \"The writeup clearly distinguishes the finite-dimensional causal estimand, infinite-dimensional nuisance functions, orthogonal score or estimator, and evaluation metrics, avoiding confusion between prediction performance and ATE estimation quality.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s02-result-writeup\", \"requirements\": \"The README or writeup describes the data-generating process, estimators, nuisance models, cross-fitting procedure, semiparametric target-vs-nuisance relationship, key numeric results, and per-hypothesis outcomes with appropriate caveats on sample size, simulation repetitions, and model misspecification.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 15, "manifest_file": "tasks/statistics/manifests/S02.yaml", "rubric_file": "tasks/statistics/rubrics/S02.json"}
+{"id": "S03", "domain": "statistics", "title": "Reliability of LLM-assisted statistical model selection", "topic": "Reliability of LLM-assisted statistical model selection under assumption violations", "domains": ["statistics", "model-selection", "llm-evaluation"], "arxiv_id": null, "venue": "ARC-Bench Statistics 2026", "metric_key": "primary_metric", "metric_direction": "maximize", "gpu_required": false, "est_wall_clock_sec": 300, "synthesis": "Large language models are increasingly used to assist with statistical data\nanalysis, including model selection, test choice, and interpretation. However,\nstatistical model selection requires matching assumptions to data-generating\nconditions, not merely producing plausible-sounding analysis text. An LLM or\nLLM-like rule-based assistant may recommend inappropriate tests when data are\nskewed, heteroskedastic, non-independent, or affected by outliers.\n\nA credible study of this topic creates a controlled benchmark of small\nstatistical-analysis tasks with known ground truth. The system should compare\nmodel-selection recommendations against an oracle or rule-based reference\nacross simulated datasets. The research question is: *can an LLM-assisted\nstatistical workflow choose appropriate statistical procedures under\nassumption violations, and how often do its recommendations lead to invalid\ninference?*", "num_hypotheses": 3, "hypotheses": "[{\"id\": \"H1\", \"statement\": \"A simple prompt-only LLM-style recommendation system is more likely to choose standard parametric tests even when assumptions are violated.\", \"measurable\": true}, {\"id\": \"H2\", \"statement\": \"Adding computed diagnostic statistics, such as skewness, variance ratio, normality tests, and sample-size information, improves statistical procedure selection accuracy.\", \"measurable\": true}, {\"id\": \"H3\", \"statement\": \"Incorrect procedure selection increases false positive rate or reduces coverage under assumption violations such as heteroskedasticity, skew, or dependence.\", \"measurable\": true}]", "experiment_design": "{\"research_question\": \"Can LLM-assisted statistical model selection reliably choose appropriate procedures when assumptions are violated?\", \"task_type\": {\"name\": \"statistical_test_selection\", \"description\": \"Given dataset summaries and analysis goals, choose an appropriate statistical test or model.\"}, \"conditions\": [{\"name\": \"text_only_recommender\", \"description\": \"Procedure selection using only natural-language task description and variable metadata.\"}, {\"name\": \"diagnostics_augmented_recommender\", \"description\": \"Procedure selection using natural-language description plus computed diagnostics.\"}, {\"name\": \"oracle_rule_based_selector\", \"description\": \"Reference selector using known data-generating process or explicit rules.\"}, {\"name\": \"always_parametric_baseline\", \"description\": \"Baseline that always chooses common parametric procedures such as t-test, ANOVA, or OLS.\"}], \"statistical_tasks\": [{\"name\": \"two_sample_mean_comparison\", \"candidate_methods\": [\"student_t_test\", \"welch_t_test\", \"mann_whitney_u\", \"permutation_test\"]}, {\"name\": \"correlation_analysis\", \"candidate_methods\": [\"pearson_correlation\", \"spearman_correlation\", \"robust_regression\"]}, {\"name\": \"linear_effect_estimation\", \"candidate_methods\": [\"ordinary_least_squares\", \"heteroskedasticity_robust_ols\", \"rank_based_method\", \"permutation_inference\"]}], \"data_generating_conditions\": [{\"name\": \"normal_equal_variance\", \"description\": \"Parametric assumptions approximately hold.\"}, {\"name\": \"normal_unequal_variance\", \"description\": \"Heteroskedastic two-sample setting.\"}, {\"name\": \"skewed_lognormal\", \"description\": \"Strongly skewed outcome distribution.\"}, {\"name\": \"outlier_contaminated\", \"description\": \"Small fraction of extreme observations.\"}, {\"name\": \"nonlinear_monotone_relationship\", \"description\": \"Correlation exists but is nonlinear or rank-based.\"}], \"metrics\": [{\"name\": \"selection_accuracy\", \"direction\": \"maximize\", \"description\": \"Fraction of tasks where selected method matches oracle-acceptable method set.\"}, {\"name\": \"false_positive_rate\", \"direction\": \"target_nominal\", \"description\": \"Empirical Type I error under null simulations.\"}, {\"name\": \"coverage\", \"direction\": \"target_0.95\", \"description\": \"Empirical confidence interval coverage when interval estimates are produced.\"}, {\"name\": \"assumption_violation_detection_rate\", \"direction\": \"maximize\", \"description\": \"Fraction of cases where the system correctly identifies relevant assumption violations.\"}, {\"name\": \"explanation_grounding_score\", \"direction\": \"maximize\", \"description\": \"Whether the written explanation cites the correct diagnostic evidence rather than generic statistical advice.\"}], \"implementation_note\": \"If external LLM API access is unavailable, the submission may simulate an\\nLLM-assisted workflow using templated recommendations, local heuristics, or\\nstored model outputs. The benchmark should focus on statistical validity of\\nrecommendations and downstream inference, not on proprietary model access.\\n\", \"simulation_settings\": {\"sample_sizes\": [30, 100, 500], \"monte_carlo_repetitions\": 500, \"seeds\": [0, 1, 2, 3, 4]}, \"expected_artifacts\": [{\"path\": \"src/generate_tasks.py\", \"description\": \"Code for generating synthetic statistical-analysis tasks.\"}, {\"path\": \"src/evaluate_selectors.py\", \"description\": \"Code for evaluating selectors and downstream inference.\"}, {\"path\": \"results/metrics.json\", \"description\": \"Machine-readable selection accuracy, false positive rate, coverage, and diagnostic-detection metrics.\"}, {\"path\": \"results/selector_decisions.csv\", \"description\": \"Machine-readable selector decisions with task metadata, diagnostics, oracle-acceptable methods, and selected procedures.\"}, {\"path\": \"results/figures/\", \"description\": \"Plots or tables comparing selectors across task types, data-generating conditions, and assumption violations.\"}, {\"path\": \"report/paper.md\", \"description\": \"Paper-style report explaining task generation, selector designs, oracle rules, diagnostic evidence, results, limitations, and per-hypothesis conclusions.\"}, {\"path\": \"README.md\", \"description\": \"Setup and run instructions for reproducing the full benchmark from a clean checkout.\"}], \"dependencies\": {\"python\": [\"numpy\", \"pandas\", \"scipy\", \"scikit-learn\", \"statsmodels\", \"matplotlib\"]}, \"compute_requirements\": {\"gpu_required\": false, \"estimated_wall_clock_sec\": 300}}", "requirements": "[{\"id\": \"req_results_json\", \"type\": \"artifact\", \"description\": \"A canonical results.json file exists at the workspace root with at least the keys: primary_metric (number), metric_key (string), metrics (object with numeric keys), hypotheses (object with h1/h2/h3 entries each carrying a `supported` boolean), summary (non-empty string).\", \"must_pass\": true}, {\"id\": \"req_selector_metrics_table\", \"type\": \"artifact\", \"description\": \"A machine-readable metrics artifact (e.g. results/metrics.json) exists reporting selection accuracy and false-positive rate (or Type I error) broken down by selector and data-generating condition, plus a results/selector_decisions.csv with per-task selector decisions and oracle-acceptable method labels.\", \"must_pass\": true}, {\"id\": \"req_selection_accuracy_numeric\", \"type\": \"numeric\", \"description\": \"results.json metrics MUST contain numeric (non-null, finite) selection accuracy values in [0, 1] for at least the text-only recommender and the diagnostics-augmented recommender, evaluated over more than one assumption-violation condition, so H1 and H2 can be evaluated.\", \"must_pass\": true}, {\"id\": \"req_hypotheses_supported_flags\", \"type\": \"discussion\", \"description\": \"results.json hypotheses.h1/h2/h3 each MUST have an explicit `supported` boolean AND a `details` string ≥ 40 characters quoting the numerical evidence (specific accuracy / false-positive-rate / coverage values + their source artifact) used to reach the verdict.\", \"must_pass\": true}, {\"id\": \"req_oracle_and_selectors\", \"type\": \"artifact\", \"description\": \"The submission implements or simulates all four selectors — text-only recommender, diagnostics-augmented recommender, oracle / rule-based reference selector, and an always-parametric baseline — and runs the selected procedures to record downstream inference (p-values, coverage, or Type I error).\", \"must_pass\": true}, {\"id\": \"req_method_validity_writeup\", \"type\": \"discussion\", \"description\": \"The report clearly distinguishes the analysis goal, data-generating condition, oracle-acceptable method set, selected method, diagnostic evidence, and downstream inference metric. Nice-to-have, not blocking.\", \"must_pass\": false}, {\"id\": \"req_llm_access_disclosed\", \"type\": \"discussion\", \"description\": \"The report states whether an external LLM API was used or whether a templated / heuristic / stored-output recommender stood in for it. Required for honest reporting but not for scientific correctness.\", \"must_pass\": false}]", "rubric": "{\"id\": \"s03-root\", \"requirements\": \"A credible controlled simulation study evaluating whether LLM-assisted or LLM-like statistical model selection chooses appropriate procedures under assumption violations. The submission should generate statistical-analysis tasks with known reference decisions, compare text-only and diagnostics-augmented recommenders against an oracle or rule-based selector, evaluate downstream inference validity, and connect the numeric findings to the three hypotheses.\", \"judging_note\": \"Score on statistical validity, reproducible task generation, appropriate reference decisions, and honest discussion of downstream inference. Do not require access to an external LLM API; a templated, heuristic, local, or stored-output recommender can receive full credit if it supports the benchmark question. Penalize generic recommendation text that is not tied to diagnostics or data-generating conditions.\", \"weight\": 1, \"sub_tasks\": [{\"id\": \"s03-code\", \"requirements\": \"The code implements a controlled benchmark for statistical procedure selection under several data-generating conditions and selector designs.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s03-code-task-generator\", \"requirements\": \"The submission generates multiple statistical-analysis tasks, including two-sample mean comparison, correlation analysis, and linear effect estimation, with known null or alternative settings and fixed random seeds.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Dataset and Model Acquisition\"}, {\"id\": \"s03-code-assumption-conditions\", \"requirements\": \"The generated tasks cover meaningful assumption regimes, including approximately normal equal-variance data and at least three violation regimes such as heteroskedasticity, skewness, outliers, nonlinear monotone relationships, or dependence.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Experimental Setup\"}, {\"id\": \"s03-code-selectors\", \"requirements\": \"The submission implements or simulates the required selectors: a text-only recommender, a diagnostics-augmented recommender, an oracle or rule-based reference selector, and a simple parametric baseline.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s03-code-diagnostics\", \"requirements\": \"The diagnostics-augmented condition computes relevant diagnostic information, such as skewness, variance ratio, normality checks, outlier indicators, sample size, monotonicity, or heteroskedasticity evidence, and makes those diagnostics available to the selector.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Method Implementation\"}, {\"id\": \"s03-code-downstream-inference\", \"requirements\": \"The code runs the selected statistical procedures or models and records downstream quantities such as p-values, confidence intervals, Type I error, coverage, or effect estimates where applicable.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Development\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s03-exec\", \"requirements\": \"Execution produces machine-readable selector and inference metrics adequate to evaluate statistical reliability.\", \"weight\": 2, \"sub_tasks\": [{\"id\": \"s03-exec-metrics\", \"requirements\": \"Execution produces a machine-readable metrics artifact, such as results/metrics.json, containing selection accuracy, false positive rate or Type I error, confidence interval coverage where applicable, assumption-violation detection rate, and explanation grounding by selector and data condition.\", \"weight\": 17.5, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s03-exec-repetitions\", \"requirements\": \"Reported metrics are aggregated over multiple simulation repetitions, seeds, task instances, or sample sizes, with enough repetition to make selector differences meaningful.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s03-exec-task-coverage\", \"requirements\": \"The executed benchmark includes more than one task type and more than one assumption-violation condition, rather than evaluating a single isolated example.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Evaluation, Metrics & Benchmarking\"}, {\"id\": \"s03-exec-artifacts\", \"requirements\": \"Execution artifacts include selector decisions, oracle-acceptable method labels, diagnostic summaries, and at least one plot or table comparing selectors across assumption regimes.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Code Execution\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}, {\"id\": \"s03-paper\", \"requirements\": \"The final paper or report addresses the three model-selection reliability hypotheses with quantitative evidence and a clear statistical narrative.\", \"weight\": 3, \"sub_tasks\": [{\"id\": \"s03-paper-h1\", \"requirements\": \"The report evaluates whether the text-only recommender over-selects standard parametric procedures under assumption violations and supports the conclusion with selector-decision metrics.\", \"weight\": 12.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s03-paper-h2\", \"requirements\": \"The report compares diagnostics-augmented and text-only selection and states whether computed diagnostics improve procedure selection accuracy or assumption-violation detection.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s03-paper-h3\", \"requirements\": \"The report analyzes whether incorrect procedure selection leads to invalid inference, such as inflated false positive rates, poor coverage, or misleading effect estimates under specific violations.\", \"weight\": 15.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s03-paper-method-validity\", \"requirements\": \"The writeup clearly distinguishes the analysis goal, data-generating condition, oracle-acceptable method set, selected method, diagnostic evidence, and downstream inference metric.\", \"weight\": 7.5, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}, {\"id\": \"s03-paper-writeup\", \"requirements\": \"The README or paper describes task generation, selector designs, diagnostic features, oracle rules, key numeric results, limitations, and per-hypothesis outcomes with appropriate caveats on simulation scope and any use or non-use of external LLM APIs.\", \"weight\": 10.0, \"sub_tasks\": [], \"task_category\": \"Result Analysis\", \"finegrained_task_category\": \"Logging, Analysis & Presentation\"}], \"task_category\": null, \"finegrained_task_category\": null}], \"task_category\": null, \"finegrained_task_category\": null}", "rubric_num_leaves": 14, "manifest_file": "tasks/statistics/manifests/S03.yaml", "rubric_file": "tasks/statistics/rubrics/S03.json"}

tasks/biology/manifests/B01.yaml

@@ -0,0 +1,164 @@
+# =============================================================================
+# B01 — E. coli succinate-production strain optimization via FBA + knockout
+# -----------------------------------------------------------------------------
+# First biology topic for ARC-Bench. Tests the full Biology-Agent integration
+# end-to-end: domain detection → biology_metabolic profile → biology prompt
+# bank → BiologyAgentSandbox at stage 12 (EXPERIMENT_RUN) → COBRApy/BIGG
+# pipeline executed inside Claude Code → flux/essentiality artifacts → judged
+# against the upgraded rubric (code/exec/results/repro).
+# =============================================================================
+
+id: B01
+title: "E. coli succinate-production strain optimization via constraint-based knockout screen"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Succinate is a top-12 platform chemical and a recurring case study in
+  metabolic engineering. The standard genome-scale model E. coli iAF1260
+  (2382 reactions, 1668 metabolites, 1261 genes) provides a well-validated
+  testbed for in-silico strain design: from a fixed glucose-minimal medium
+  one can predict (a) the maximum aerobic biomass growth rate, (b) the
+  maximum theoretical succinate secretion flux once growth is held at a feasible
+  fraction of optimum, (c) the set of single- and double-gene knockouts
+  that decouple biomass from succinate secretion enough to push the strain
+  toward a high-secretion non-growth-coupled phenotype.
+
+  A credible CPU-scale study of this topic (a) loads iAF1260 from BIGG and
+  validates aerobic biomass > 0.7 1/h on glucose-minimal medium, (b) runs
+  parsimonious FBA (pFBA) at the wild-type optimum and reports the central
+  carbon flux distribution, (c) sweeps the production envelope (biomass vs
+  succinate secretion) using flux variability analysis (FVA) on the
+  succinate exchange across a grid of biomass-fraction-of-optimum values,
+  (d) runs a single-gene knockout screen (≤100 candidate genes drawn from
+  central carbon metabolism, fermentation, and TCA cycle) and ranks each KO
+  by post-KO succinate secretion flux at 50% of WT growth, (e) reports the top three
+  KO strains with mechanistic explanation tied to the network topology.
+
+  The research question is: *which single-gene knockouts of central carbon
+  metabolism in E. coli iAF1260 produce the largest predicted succinate
+  secretion flux while preserving at least 50% of wild-type biomass growth, and what
+  is the mechanistic role of each in shifting flux toward succinate?*
+
+hypotheses:
+  - id: H1
+    statement: "Wild-type aerobic E. coli iAF1260 on glucose-minimal medium predicts a biomass growth rate within ±10% of 0.736 1/h (the published BIGG iAF1260 reference value)."
+    measurable: true
+  - id: H2
+    statement: "The biomass-vs-succinate production envelope, computed via FVA at biomass fractions from 0.1 to 1.0 of WT optimum, is monotonically non-increasing in succinate as biomass approaches the WT optimum (i.e., succinate is growth-competing under aerobic glucose conditions)."
+    measurable: true
+  - id: H3
+    statement: "At least 3 of the top-5 single-gene knockouts identified by the screen as maximizing succinate secretion flux at 50% WT-growth disrupt canonical competing by-product or respiration pathways described in the metabolic-engineering literature (e.g., pyruvate-formate-lyase pflB, lactate dehydrogenase ldhA, alcohol dehydrogenase adhE, acetate kinase ackA, phosphate acetyltransferase pta, or their isozymes), or are accompanied by a mechanistic explanation for why the knockout redirects flux toward succinate under the chosen medium."
+    measurable: true
+
+experiment_design:
+  research_question: "Which single-gene knockouts of central carbon metabolism in E. coli iAF1260 produce the largest predicted succinate secretion flux while preserving at least 50% of wild-type biomass growth, and what is the mechanistic role of each in shifting flux toward succinate?"
+  conditions:
+    - name: "wt_aerobic_glucose"
+      description: "Wild-type iAF1260 on glucose-minimal medium with O2 unconstrained. Reference condition."
+    - name: "wt_aerobic_glucose_pfba"
+      description: "Same as wt_aerobic_glucose but with pFBA (parsimonious FBA) for unique flux distribution."
+    - name: "production_envelope_succinate"
+      description: "FVA on succinate exchange across biomass-fraction-of-optimum grid {0.1, 0.2, ..., 1.0}."
+    - name: "single_ko_screen_central_carbon"
+      description: "Single-gene knockouts on a curated set of ≤100 central-carbon / fermentation / TCA genes (glycolysis, PPP, TCA, anaplerotic, fermentation by-products); for each, FBA at fixed biomass = 0.5 × WT optimum; record succinate exchange flux."
+  baselines:
+    - "wt_aerobic_glucose (no perturbation) is the no-engineering baseline"
+    - "Published BIGG iAF1260 reference growth rate (0.736 1/h on glucose-minimal aerobic) — the model-validity gate"
+  metrics:
+    - name: "wt_growth_rate"
+      direction: "match_reference"
+      description: "Wild-type biomass exchange flux (1/h) on glucose-minimal aerobic; reference value 0.736."
+    - name: "max_succinate_flux_mmol_gDW_h_at_zero_growth"
+      direction: "maximize"
+      description: "Succinate exchange flux (mmol/gDW/h) at biomass = 0 (theoretical maximum)."
+    - name: "succinate_flux_mmol_gDW_h_at_half_growth"
+      direction: "maximize"
+      description: "Succinate exchange flux at biomass = 0.5 × WT optimum, wild-type strain."
+    - name: "best_ko_succinate_flux_mmol_gDW_h"
+      direction: "maximize"
+      description: "Maximum succinate exchange flux over the single-KO screen at biomass = 0.5 × WT optimum."
+  datasets:
+    - name: "iAF1260"
+      source: "BIGG database (http://bigg.ucsd.edu/models/iAF1260) via cobra.io.load_model('iAF1260')"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 1800
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B01.json"
+
+# ---------------------------------------------------------------------------
+# Agent-mode requirements (consumed by researchclaw.pipeline.requirements_judge
+# at stage 15 RESEARCH_DECISION).  Each item:
+#   id           — stable string identifier
+#   type         — advisory: numeric | discussion | artifact (LLM uses freely)
+#   description  — natural-language statement of what the agent must produce
+#   must_pass    — true → unmet ⇒ rerun (1 retry max); false → optional
+#
+# These are a separate semantic layer from the rubric: rubrics score quality,
+# requirements gate proceed-vs-rerun.  Keep the must_pass set tight (2–4 items)
+# so the agent has an unambiguous bar to clear.
+# ---------------------------------------------------------------------------
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: >-
+      A canonical results.json file exists at the workspace root with at least
+      the keys: primary_metric (number), metrics (object of numeric keys),
+      hypotheses (object with h1/h2/h3 entries each carrying a `supported`
+      boolean), summary (non-empty string).
+    must_pass: true
+
+  - id: req_wt_growth_value
+    type: numeric
+    description: >-
+      results.json metrics MUST contain a numeric value for
+      `wt_growth_observed_1_per_h` (or an equivalent wt_growth_*_1_per_h key)
+      that is non-null and finite. The wild-type growth rate must be reported,
+      otherwise H1 cannot be evaluated.
+    must_pass: true
+
+  - id: req_top5_ko
+    type: discussion
+    description: >-
+      results.json structured_results MUST contain a list of the top 5 ranked
+      single-gene knockouts with at least {gene_id, gene_name,
+      ko_succinate_flux_mmol_gDW_h,
+      is_canonical} fields each. The list need not include the textbook
+      canonical genes — but it MUST exist and identify at least 5 candidates.
+    must_pass: true
+
+  - id: req_envelope_figure
+    type: artifact
+    description: >-
+      A production-envelope figure (biomass-vs-succinate, or analogous) must
+      exist under figures/ or analysis/ in PDF or PNG format with axes labeled
+      and units shown. Either a single multi-curve figure or per-condition
+      figures both qualify.
+    must_pass: true
+
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >-
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: >-
+      Top-3 KO targets are discussed mechanistically — naming the pathway each
+      KO disables and a one-sentence rationale for why the disruption
+      redirects flux to succinate.  Acceptable inside summary, structured_results,
+      or a separate writeup field.
+    must_pass: false  # Optional: nice-to-have, not blocking
+
+  - id: req_seed_documented
+    type: discussion
+    description: >-
+      The solver backend (cobra optlang interface name + version) and any RNG
+      seeds used (e.g. for sampling) are reported in results.json. Required for
+      reproducibility but not for scientific correctness.
+    must_pass: false  # Optional

tasks/biology/manifests/B02.yaml

@@ -0,0 +1,100 @@
+id: B02
+title: "E. coli acetate overflow metabolism phase map under glucose and oxygen limitation"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Overflow acetate secretion in E. coli is a canonical constraint-based
+  modelling phenotype. A credible CPU-scale study loads a validated E. coli
+  genome-scale model, fixes a glucose-minimal medium, sweeps glucose and oxygen
+  uptake bounds, computes growth and major secretion products, and identifies
+  the transition from respiratory growth to overflow/fermentative by-product
+  secretion.
+
+  The research question is: under which glucose and oxygen uptake regimes does
+  E. coli switch from primarily respiratory growth to acetate-secreting overflow
+  metabolism, and how robust is that regime boundary across FBA and pFBA?
+
+hypotheses:
+  - id: H1
+    statement: "The selected E. coli BIGG model grows on aerobic glucose-minimal medium with a positive biomass objective and no infeasible solver status."
+    measurable: true
+  - id: H2
+    statement: "A two-dimensional glucose-vs-oxygen phase plane contains distinct regimes with high oxygen/low acetate and low oxygen/high acetate secretion."
+    measurable: true
+  - id: H3
+    statement: "pFBA reduces total internal flux while preserving the same growth-rate regime boundaries to within a small tolerance."
+    measurable: true
+
+experiment_design:
+  research_question: "Under which glucose and oxygen uptake regimes does E. coli switch from respiratory growth to acetate overflow?"
+  conditions:
+    - name: "aerobic_glucose_reference"
+      description: "Wild-type E. coli iJO1366 or iAF1260 on glucose-minimal medium with oxygen available."
+    - name: "glucose_oxygen_phase_plane"
+      description: "Grid over glucose uptake and oxygen uptake bounds; compute biomass, acetate, ethanol, lactate, succinate, and CO2 exchange fluxes."
+    - name: "pfba_phase_plane_check"
+      description: "Run pFBA on the same grid or a representative subset and compare growth plus secretion regimes."
+  baselines:
+    - "High-oxygen glucose condition is the respiratory baseline."
+    - "Zero-oxygen glucose condition is the fermentative baseline."
+  metrics:
+    - name: "growth_rate_1_per_h"
+      direction: "maximize"
+      description: "Biomass objective flux for each nutrient grid point."
+    - name: "acetate_flux_mmol_gDW_h"
+      direction: "characterize"
+      description: "Acetate exchange secretion flux."
+    - name: "overflow_boundary_o2_bound"
+      direction: "estimate"
+      description: "Approximate oxygen bound below which acetate secretion exceeds a chosen threshold."
+    - name: "pfba_total_flux_norm"
+      direction: "minimize"
+      description: "Total absolute flux under pFBA."
+  datasets:
+    - name: "E. coli GSMM"
+      source: "BIGG iJO1366 or iAF1260 via cobra.io.load_model"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 1800
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B02.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "A results.json file exists with primary_metric, metrics, hypotheses, summary, and structured_results."
+    must_pass: true
+  - id: req_phase_plane_table
+    type: artifact
+    description: "A machine-readable phase-plane table exists with glucose_bound, oxygen_bound, growth_rate, and acetate_flux columns."
+    must_pass: true
+  - id: req_phase_plane_figure
+    type: artifact
+    description: "A PNG or PDF heatmap/phase-plane figure exists with glucose and oxygen axes and units."
+    must_pass: true
+  - id: req_aerobic_ref_growth
+    type: numeric
+    description: >-
+      results.json metrics MUST contain a numeric value for
+      `aerobic_glucose_growth_rate_1_per_h` (or an equivalent aerobic_growth_*_1_per_h key)
+      that is non-null, finite, and positive. The aerobic reference growth rate must be
+      reported, otherwise H1 cannot be evaluated.
+    must_pass: true
+
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "The overflow boundary and its mechanistic basis are discussed — why E. coli switches to acetate-secreting overflow at the identified nutrient threshold. Acceptable inside summary, phase_plane_interpretation, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/manifests/B03.yaml

@@ -0,0 +1,90 @@
+id: B03
+title: "Anaerobic E. coli lactate-overproduction knockout screen"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Lactate production under anaerobic glucose conditions is a tractable
+  metabolic-engineering task for constraint-based modelling. A credible study
+  loads an E. coli BIGG model, validates anaerobic growth, computes WT lactate
+  secretion, screens central-carbon and fermentation knockouts, and ranks
+  interventions that increase lactate secretion while preserving growth.
+
+hypotheses:
+  - id: H1
+    statement: "The model supports positive anaerobic glucose growth after oxygen uptake is closed."
+    measurable: true
+  - id: H2
+    statement: "The lactate production envelope shows a tradeoff between biomass growth and maximum lactate secretion."
+    measurable: true
+  - id: H3
+    statement: "Top lactate-improving knockouts disable competing ethanol, formate, acetate, or succinate by-product routes, or have a specific mechanistic explanation."
+    measurable: true
+
+experiment_design:
+  research_question: "Which single-gene knockouts increase anaerobic lactate secretion in E. coli while preserving at least 30-50% of WT growth?"
+  conditions:
+    - name: "wt_anaerobic_glucose"
+      description: "E. coli iJO1366 or iAF1260 on glucose-minimal medium with oxygen uptake closed."
+    - name: "lactate_production_envelope"
+      description: "Sweep biomass fraction and optimize lactate secretion."
+    - name: "single_ko_screen_fermentation"
+      description: "Screen central-carbon and fermentation genes; rank by lactate flux at fixed growth fraction."
+  baselines:
+    - "Wild-type anaerobic glucose lactate secretion."
+    - "No-knockout lactate envelope."
+  metrics:
+    - name: "anaerobic_wt_growth_1_per_h"
+      direction: "maximize"
+      description: "WT biomass flux with oxygen uptake set to zero."
+    - name: "lactate_flux_mmol_gDW_h"
+      direction: "maximize"
+      description: "Lactate exchange secretion flux."
+    - name: "lactate_yield_per_glucose"
+      direction: "maximize"
+      description: "Lactate secretion divided by glucose uptake."
+    - name: "growth_fraction"
+      direction: "threshold"
+      description: "KO biomass relative to WT anaerobic growth."
+  datasets:
+    - name: "E. coli GSMM"
+      source: "BIGG iJO1366 or iAF1260 via cobra.io.load_model"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 1800
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B03.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "results.json contains primary_metric, metrics, hypotheses, summary, and top KO structured results."
+    must_pass: true
+  - id: req_anaerobic_medium
+    type: numeric
+    description: "Oxygen uptake is explicitly closed and anaerobic WT growth is reported."
+    must_pass: true
+  - id: req_top5_ko
+    type: discussion
+    description: "Top 5 lactate-ranked knockouts are reported with gene_id, gene_name, growth_fraction, lactate_flux, and mechanism/canonical flag."
+    must_pass: true
+  - id: req_envelope_figure
+    type: artifact
+    description: "A biomass-vs-lactate production envelope figure exists with labelled axes and units."
+    must_pass: true
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "Top-3 lactate-improving KO targets are discussed mechanistically — naming the pathway each KO disables and a one-sentence rationale for why the disruption redirects anaerobic flux to lactate. Acceptable inside summary, structured_results, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/manifests/B04.yaml

@@ -0,0 +1,87 @@
+id: B04
+title: "S. cerevisiae ethanol yield under oxygen limitation and carbon-source changes"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Yeast ethanol fermentation is a natural extension beyond E. coli while still
+  staying inside BIGG/COBRApy constraint-based modelling. A credible study
+  loads iMM904 or a comparable yeast GSMM, validates growth on glucose, sweeps
+  oxygen availability, compares glucose and alternative carbon sources, and
+  quantifies ethanol secretion and yield.
+
+hypotheses:
+  - id: H1
+    statement: "The yeast model grows on glucose-minimal medium and produces a feasible FBA solution."
+    measurable: true
+  - id: H2
+    statement: "Ethanol secretion/yield increases as oxygen uptake is restricted relative to fully aerobic growth."
+    measurable: true
+  - id: H3
+    statement: "Carbon-source swaps produce distinct growth and ethanol-yield profiles, with glucose supporting stronger fermentation than at least one alternative carbon source."
+    measurable: true
+
+experiment_design:
+  research_question: "How do oxygen limitation and carbon source alter predicted ethanol secretion and yield in S. cerevisiae?"
+  conditions:
+    - name: "yeast_glucose_reference"
+      description: "S. cerevisiae iMM904 or comparable model on glucose-minimal medium."
+    - name: "oxygen_sweep"
+      description: "Sweep oxygen uptake from anaerobic/low oxygen to aerobic while glucose uptake is fixed."
+    - name: "carbon_source_swap"
+      description: "Compare glucose, fructose, galactose, glycerol, and acetate when supported by the model."
+  baselines:
+    - "Aerobic glucose condition."
+    - "Anaerobic or oxygen-limited glucose condition."
+  metrics:
+    - name: "growth_rate_1_per_h"
+      direction: "maximize"
+      description: "Biomass flux."
+    - name: "ethanol_flux_mmol_gDW_h"
+      direction: "maximize"
+      description: "Ethanol exchange secretion flux."
+    - name: "ethanol_yield_per_carbon_uptake"
+      direction: "maximize"
+      description: "Ethanol secretion divided by substrate uptake."
+  datasets:
+    - name: "S. cerevisiae iMM904"
+      source: "BIGG iMM904 via cobra.io.load_model if available, or documented local SBML/JSON yeast model."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 1800
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B04.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "results.json contains yeast model ID, metrics, hypotheses, and structured oxygen/carbon-source results."
+    must_pass: true
+  - id: req_oxygen_sweep
+    type: artifact
+    description: "A machine-readable oxygen-sweep table exists with oxygen bound, growth, ethanol flux, and yield."
+    must_pass: true
+  - id: req_carbon_source_table
+    type: artifact
+    description: "A carbon-source comparison table exists, including unsupported sources marked explicitly."
+    must_pass: true
+  - id: req_figure
+    type: artifact
+    description: "At least one oxygen-vs-ethanol or carbon-source yield figure exists with units."
+    must_pass: true
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "The ethanol-yield response to oxygen and carbon source is discussed mechanistically — why oxygen restriction increases ethanol yield and how carbon-source metabolism shapes fermentation. Acceptable inside summary, structured_results, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/manifests/B05.yaml

@@ -0,0 +1,90 @@
+id: B05
+title: "M. tuberculosis condition-specific essentiality and metabolic drug-target prioritisation"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Constraint-based essentiality analysis can prioritise metabolic drug-target
+  hypotheses in pathogen models. A credible study loads an M. tuberculosis GSMM
+  such as iNJ661 when available, validates biomass production under a documented
+  medium, performs single-gene and/or single-reaction deletion, and ranks
+  essential targets by growth impact and subsystem interpretability.
+
+hypotheses:
+  - id: H1
+    statement: "The selected M. tuberculosis model produces positive biomass under the documented reference medium."
+    measurable: true
+  - id: H2
+    statement: "Single-gene or single-reaction deletion identifies a non-empty set of essential metabolic targets under the reference condition."
+    measurable: true
+  - id: H3
+    statement: "Prioritised targets are enriched in interpretable core metabolic subsystems such as cell-wall precursor, cofactor, lipid, energy, or amino-acid metabolism."
+    measurable: true
+
+experiment_design:
+  research_question: "Which condition-specific metabolic genes or reactions are predicted essential in M. tuberculosis and are plausible drug-target hypotheses?"
+  conditions:
+    - name: "mtb_reference_medium"
+      description: "M. tuberculosis iNJ661 or comparable model under a documented reference medium."
+    - name: "single_gene_deletion"
+      description: "Single-gene deletion screen when GPR rules are available."
+    - name: "single_reaction_deletion"
+      description: "Single-reaction deletion screen as a fallback or complementary target set."
+  baselines:
+    - "Wild-type reference growth."
+    - "Nonessential deletion growth distribution."
+  metrics:
+    - name: "wt_growth_rate_1_per_h"
+      direction: "maximize"
+      description: "Reference biomass flux."
+    - name: "growth_fraction_after_deletion"
+      direction: "minimize"
+      description: "Deletion biomass divided by WT biomass."
+    - name: "essential_target_count"
+      direction: "characterize"
+      description: "Count of genes/reactions below essentiality threshold."
+    - name: "subsystem_enrichment"
+      direction: "characterize"
+      description: "Subsystem distribution among essential targets."
+  datasets:
+    - name: "M. tuberculosis GSMM"
+      source: "BIGG iNJ661 or documented local SBML/JSON model."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 2400
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B05.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "results.json contains model ID, WT growth, essential target counts, top targets, and hypothesis verdicts."
+    must_pass: true
+  - id: req_deletion_table
+    type: artifact
+    description: "A gene or reaction deletion table exists with target ID, status, growth, growth_fraction, and essential flag."
+    must_pass: true
+  - id: req_top_targets
+    type: discussion
+    description: "Top 10 prioritised essential targets are reported with subsystem/pathway and rationale."
+    must_pass: true
+  - id: req_figure
+    type: artifact
+    description: "At least one essentiality distribution or subsystem-enrichment figure exists."
+    must_pass: true
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "Top-3 essential target hypotheses are framed mechanistically — naming the metabolic subsystem each target belongs to and why its disruption is predicted lethal under the reference medium. Claims must be framed as computational hypotheses, not validated drugs. Acceptable inside summary, structured_results, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/manifests/B06.yaml

@@ -0,0 +1,85 @@
+id: B06
+title: "E. coli carbon-source robustness and condition-dependent essential genes"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  Carbon-source swaps are a direct mfa-agent capability and produce interpretable
+  condition-dependent phenotypes. A credible study loads an E. coli GSMM, defines
+  a base minimal medium, tests multiple single-carbon sources, runs essentiality
+  analysis on a focused gene set under each feasible source, and identifies genes
+  whose essentiality changes with carbon source.
+
+hypotheses:
+  - id: H1
+    statement: "The model predicts positive growth on glucose and at least two additional supported carbon sources."
+    measurable: true
+  - id: H2
+    statement: "Growth rates and secretion profiles differ substantially across carbon sources."
+    measurable: true
+  - id: H3
+    statement: "A focused central-carbon gene set contains condition-dependent essential genes whose deletion effects vary by carbon source."
+    measurable: true
+
+experiment_design:
+  research_question: "Which E. coli metabolic vulnerabilities are carbon-source dependent across glucose, glycerol, acetate, succinate, and fructose-like conditions?"
+  conditions:
+    - name: "carbon_source_panel"
+      description: "Close background carbon uptake, open one carbon source at a time, and run FBA/pFBA."
+    - name: "focused_gene_essentiality_by_carbon"
+      description: "For feasible carbon sources, delete a focused set of central-carbon genes and record growth fractions."
+  baselines:
+    - "Glucose-minimal medium."
+    - "Wild-type growth for each carbon source."
+  metrics:
+    - name: "growth_rate_by_carbon"
+      direction: "characterize"
+      description: "WT biomass flux per carbon source."
+    - name: "major_secretions_by_carbon"
+      direction: "characterize"
+      description: "Positive exchange fluxes under pFBA."
+    - name: "condition_dependent_essential_count"
+      direction: "maximize"
+      description: "Genes essential under one source but not another."
+  datasets:
+    - name: "E. coli GSMM"
+      source: "BIGG iJO1366 or iAF1260 via cobra.io.load_model"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 2400
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B06.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "results.json contains carbon-source growth table summary, essentiality counts, hypothesis verdicts, and summary."
+    must_pass: true
+  - id: req_carbon_table
+    type: artifact
+    description: "A carbon-source comparison table exists with source, exchange reaction, status, growth, and secretion profile fields."
+    must_pass: true
+  - id: req_essentiality_matrix
+    type: artifact
+    description: "A gene-by-carbon essentiality/growth-fraction matrix exists for feasible carbon sources."
+    must_pass: true
+  - id: req_heatmap
+    type: artifact
+    description: "A heatmap or clustered plot of condition-dependent gene essentiality exists with labelled axes."
+    must_pass: true
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "Top-3 condition-dependent essential genes are explained mechanistically — why each gene is essential under one carbon source but not another, referencing the metabolic pathway context. Acceptable inside summary, structured_results, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/manifests/B07.yaml

@@ -0,0 +1,92 @@
+id: B07
+title: "Reproducible FBA protocol benchmark: FBA vs pFBA vs loopless/FVA on E. coli"
+arxiv_id: null
+venue: "ARC-Bench Biology 2026"
+paper_asset: null
+
+synthesis: |
+  A method benchmark is well matched to mfa-agent because it tests robust use of
+  COBRApy APIs rather than biological novelty alone. A credible study loads a
+  validated E. coli GSMM, runs standard FBA, pFBA, loopless FBA when available,
+  and FVA at a fixed fraction of optimum, then compares growth, flux sparsity,
+  runtime, and reaction-level variability.
+
+hypotheses:
+  - id: H1
+    statement: "FBA, pFBA, and loopless FBA preserve the same biomass optimum within solver tolerance under the same medium."
+    measurable: true
+  - id: H2
+    statement: "pFBA produces a lower total absolute flux norm and/or fewer active reactions than unconstrained FBA."
+    measurable: true
+  - id: H3
+    statement: "FVA reveals that only a subset of central-carbon reactions are tightly constrained near optimum while others remain variable."
+    measurable: true
+
+experiment_design:
+  research_question: "How do standard FBA, pFBA, loopless FBA, and FVA differ in flux parsimony and variability while preserving E. coli growth predictions?"
+  conditions:
+    - name: "standard_fba"
+      description: "Maximize biomass on aerobic glucose-minimal medium."
+    - name: "pfba"
+      description: "Maximize biomass then minimize total absolute flux."
+    - name: "loopless_fba"
+      description: "Run loopless solution when available and compare objective/fluxes."
+    - name: "fva_95_percent_growth"
+      description: "Run FVA at 95% of optimum and classify rigid vs flexible reactions."
+  baselines:
+    - "Standard FBA biomass optimum."
+    - "COBRApy solver status and runtime."
+  metrics:
+    - name: "growth_rate_1_per_h"
+      direction: "match"
+      description: "Biomass flux for each method."
+    - name: "total_abs_flux"
+      direction: "minimize"
+      description: "Sum of absolute reaction fluxes."
+    - name: "active_reaction_count"
+      direction: "minimize"
+      description: "Number of reactions with absolute flux above tolerance."
+    - name: "fva_width"
+      direction: "characterize"
+      description: "Maximum minus minimum FVA interval per reaction."
+  datasets:
+    - name: "E. coli GSMM"
+      source: "BIGG iJO1366 or iAF1260 via cobra.io.load_model"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 1800
+
+rubric_path: "experiments/arc_bench/config/biology/rubrics/B07.json"
+
+requirements:
+  - id: req_results_json
+    type: artifact
+    description: "results.json contains method-level metrics, hypotheses, summary, and solver/version metadata."
+    must_pass: true
+  - id: req_method_table
+    type: artifact
+    description: "A method-comparison table exists with method, status, growth, total_abs_flux, active_reactions, and runtime."
+    must_pass: true
+  - id: req_fva_table
+    type: artifact
+    description: "An FVA table exists with reaction ID, minimum, maximum, width, and rigid/flexible classification."
+    must_pass: true
+  - id: req_figures
+    type: artifact
+    description: "At least one labelled figure compares methods or FVA variability."
+    must_pass: true
+  - id: req_h1_h2_h3_supported_flags
+    type: discussion
+    description: >
+      Each of hypotheses.h1, hypotheses.h2, hypotheses.h3 in results.json MUST
+      have an explicit `supported` boolean and a `details` string of at least
+      40 characters describing the evidence used to reach that verdict.
+    must_pass: true
+  - id: req_mechanistic_writeup
+    type: discussion
+    description: "The protocol comparison is summarised with a method recommendation — which method is most reproducible and why, referencing flux-norm and FVA evidence. Acceptable inside summary, method_comparison, or a separate writeup field."
+    must_pass: false  # Optional: nice-to-have, not blocking
+  - id: req_seed_documented
+    type: discussion
+    description: "The solver backend (cobra optlang interface name + version) and COBRApy version are reported in results.json. Required for reproducibility but not for scientific correctness."
+    must_pass: false  # Optional

tasks/biology/rubrics/B01.json

@@ -0,0 +1,166 @@
+{
+    "id": "B01",
+    "requirements": "A credible constraint-based experiment on E. coli iAF1260 that (a) loads the BIGG model and validates wild-type growth, (b) computes a flux distribution under the standard objective, (c) characterises the biomass-vs-succinate production envelope, (d) runs a single-gene knockout screen targeted at central carbon metabolism, and (e) ties the top-ranked knockouts to the network mechanism. Partial but well-motivated evidence deserves partial credit; rigid wording or naming should not penalize a substantively correct experiment.",
+    "judging_note": "Score on biological substance and quantitative correctness, not exact threshold satisfaction. Reproducing canonical FBA numbers (WT growth ~0.7, well-known knockouts like pflB/ldhA/adhE/ackA-pta) is strong evidence; novel-but-coherent design choices are acceptable.",
+    "weight": 1,
+    "sub_tasks": [
+        {
+            "id": "b01-code",
+            "requirements": "The constraint-based pipeline is implemented with COBRApy and a BIGG genome-scale model.",
+            "weight": 2,
+            "sub_tasks": [
+                {
+                    "id": "b01-code-model",
+                    "requirements": "The submission loads E. coli iAF1260 (or a comparable validated genome-scale metabolic model) from BIGG via COBRApy, with the medium constraints (glucose uptake, O2 availability) explicitly set and the biomass objective explicitly named.",
+                    "weight": 8.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Method Implementation"
+                },
+                {
+                    "id": "b01-code-fba",
+                    "requirements": "FBA and pFBA are invoked correctly to obtain wild-type growth and a unique flux distribution; output flux table covers exchange + central-carbon reactions at minimum.",
+                    "weight": 6.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Method Implementation"
+                },
+                {
+                    "id": "b01-code-envelope",
+                    "requirements": "The biomass-vs-succinate production envelope is computed by sweeping a fraction-of-optimum biomass constraint and running FVA (or two-step FBA) on the succinate exchange at each grid point.",
+                    "weight": 4.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Experimental Setup"
+                },
+                {
+                    "id": "b01-code-koscreen",
+                    "requirements": "The single-gene knockout screen iterates over a curated central-carbon / fermentation / TCA gene set (≤100), applies model.genes.<g>.knock_out() (or equivalent reaction-bound deletion), and records succinate flux at fixed biomass = 0.5 × WT optimum.",
+                    "weight": 4.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Method Implementation"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "b01-exec",
+            "requirements": "Execution produces the FBA / FVA / KO numbers needed to evaluate the hypotheses without crashing.",
+            "weight": 2,
+            "sub_tasks": [
+                {
+                    "id": "b01-exec-validity",
+                    "requirements": "Solver returns optimal status for the wild-type FBA and at least 90% of the KO conditions; no negative biomass or non-physical fluxes (succinate secretion negative under aerobic glucose, etc.) appear without explanation.",
+                    "weight": 8.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Execution",
+                    "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+                },
+                {
+                    "id": "b01-exec-physics",
+                    "requirements": "Mass balance, charge balance, and biomass-positive checks pass for the loaded model; this is the biology-validity gate analogous to physics-validity (gauge invariance, unitarity) in the physics rubric.",
+                    "weight": 8.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Execution",
+                    "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+                },
+                {
+                    "id": "b01-exec-results",
+                    "requirements": "A machine-readable results artifact (results.json or simulations/*.csv) records WT growth rate, the production-envelope grid (biomass fraction → succinate secretion flux in mmol/gDW/h), and a per-KO succinate-flux table. If yield is reported, define it separately as succinate secretion flux divided by glucose uptake flux.",
+                    "weight": 6.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Execution",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "b01-results",
+            "requirements": "The results analysis addresses the three hypotheses with quantitative evidence and a clear narrative tied to network mechanism.",
+            "weight": 3,
+            "sub_tasks": [
+                {
+                    "id": "b01-result-h1-quant",
+                    "requirements": "Quantitative test of H1: predicted WT aerobic growth rate is reported and compared to the published BIGG reference 0.736 1/h. Score 100% if relative error < 3%, 67% if < 10%, 33% if < 30%, 0% otherwise.",
+                    "weight": 12.0,
+                    "sub_tasks": [],
+                    "task_category": "Result Analysis",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                },
+                {
+                    "id": "b01-result-h2-quant",
+                    "requirements": "Quantitative test of H2: production envelope shows succinate is growth-competing — succinate flux is non-increasing as biomass approaches WT optimum. Score 100% if monotonic over ≥8 of 10 grid points, 67% if ≥6 of 10, 33% if ≥4 of 10.",
+                    "weight": 10.0,
+                    "sub_tasks": [],
+                    "task_category": "Result Analysis",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                },
+                {
+                    "id": "b01-result-h3-quant",
+                    "requirements": "Quantitative test of H3: among the top-5 KO ranked by succinate secretion flux at 0.5 × WT growth, count how many disrupt canonical competing by-product or respiration pathways for succinate-overproduction logic, such as {pflB, ldhA, adhE, ackA, pta, acetate/ethanol/lactate branch competitors, oxygen-respiration competitors}, or include a specific mechanistic explanation for redirecting flux toward succinate under the chosen medium. Score 100% if ≥3 of 5 meet this canonical-or-mechanistic criterion, 67% if 2 of 5, 33% if 1 of 5, 0% otherwise.",
+                    "weight": 10.0,
+                    "sub_tasks": [],
+                    "task_category": "Result Analysis",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                },
+                {
+                    "id": "b01-result-figure",
+                    "requirements": "At least one publication-quality figure is produced — typically the production envelope (biomass on x, succinate secretion flux on y, with WT and best KO overlaid) and/or the KO essentiality / flux heatmap. Axes labeled with units (1/h, mmol/gDW/h), legend present.",
+                    "weight": 8.0,
+                    "sub_tasks": [],
+                    "task_category": "Result Analysis",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                },
+                {
+                    "id": "b01-result-writeup",
+                    "requirements": "The README or results writeup discusses each hypothesis outcome, names the top knockouts with their gene/enzyme identity, and gives a one-paragraph mechanistic interpretation (which competing pathways the KOs disable, why this redirects flux to succinate).",
+                    "weight": 6.0,
+                    "sub_tasks": [],
+                    "task_category": "Result Analysis",
+                    "finegrained_task_category": "Logging, Analysis & Presentation"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "b01-repro",
+            "requirements": "Reproducibility: the model, run cards, and seeds are checked into artifacts so a fresh clone can rerun and obtain matching numbers.",
+            "weight": 2,
+            "sub_tasks": [
+                {
+                    "id": "b01-repro-model",
+                    "requirements": "The loaded GSMM is either persisted (models/iAF1260.json) or pinned by version + BIGG URL in the writeup; a downstream user can locate the exact model object used.",
+                    "weight": 5.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Reproducibility"
+                },
+                {
+                    "id": "b01-repro-runcards",
+                    "requirements": "The medium definition (exchange bounds), objective function, FVA fraction-of-optimum, and KO gene list are saved to a config / params file (not only inline in code).",
+                    "weight": 5.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Reproducibility"
+                },
+                {
+                    "id": "b01-repro-seeds",
+                    "requirements": "Solver settings (LP backend name + version, tolerance) are recorded; if any sampling step is used, the RNG seed is logged. A second run reproduces the central WT growth rate to within solver tolerance.",
+                    "weight": 5.0,
+                    "sub_tasks": [],
+                    "task_category": "Code Development",
+                    "finegrained_task_category": "Reproducibility"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        }
+    ],
+    "task_category": null,
+    "finegrained_task_category": null
+}

tasks/biology/rubrics/B02.json

@@ -0,0 +1,166 @@
+{
+  "id": "B02",
+  "requirements": "A credible constraint-based E. coli acetate-overflow phase-plane study that loads a validated BIGG model, sets glucose and oxygen medium bounds, sweeps a two-dimensional nutrient grid, reports growth and by-product secretion, and interprets the acetate overflow boundary.",
+  "judging_note": "Score biological substance and quantitative reproducibility over exact thresholds. iJO1366, iAF1260, or a comparable curated E. coli GSMM are acceptable.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b02-code",
+      "requirements": "The COBRApy implementation defines the model, medium, phase-plane grid, FBA/pFBA runs, and secretion extraction.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b02-code-model",
+          "requirements": "Loads a validated E. coli BIGG model, explicitly sets glucose and oxygen exchange bounds, and names the biomass objective.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b02-code-grid",
+          "requirements": "Constructs a two-dimensional glucose-by-oxygen grid with at least 8 values per axis and restores model state between conditions.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b02-code-products",
+          "requirements": "Extracts biomass plus acetate, ethanol, lactate, succinate, CO2, and O2/glucose exchange fluxes where present.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b02-code-pfba",
+          "requirements": "Runs pFBA on the full grid or a justified representative subset and records total absolute flux.",
+          "weight": 4,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b02-exec",
+      "requirements": "Execution produces stable phase-plane artifacts without infeasible or missing-condition failures dominating the study.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b02-exec-status",
+          "requirements": "At least 90% of grid points solve to optimal or biologically expected zero-growth status with explicit handling.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b02-exec-physics",
+          "requirements": "Mass-balance check passes for the loaded model; no non-physical secretion fluxes appear under aerobic conditions without explicit justification (e.g. acetate secretion is zero under high-oxygen/low-glucose conditions in the reference FBA); biomass flux is positive under aerobic glucose reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b02-exec-artifacts",
+          "requirements": "Writes results.json and a CSV/TSV phase-plane table with nutrient bounds, growth, and secretion fluxes.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b02-results",
+      "requirements": "The analysis identifies and explains acetate-overflow regimes with figures and quantitative criteria.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b02-result-h1",
+          "requirements": "Reports reference aerobic glucose growth and confirms model feasibility and positive biomass production. Score 100% if a positive finite biomass flux is reported for aerobic glucose with optimal solver status; 67% if growth is positive but solver status is not verified; 33% if only feasibility is stated without a numeric growth value; 0% otherwise.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b02-result-h2",
+          "requirements": "Quantifies an acetate secretion boundary or threshold across oxygen/glucose regimes and distinguishes respiratory vs overflow conditions. Score 100% if an acetate-overflow boundary is quantified (O2 or glucose uptake threshold stated) with ≥2 distinct secretion regimes visible in the phase-plane table; 67% if overflow is identified without a quantified boundary; 33% if only one secretion regime is described; 0% otherwise.",
+          "weight": 12,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b02-result-h3",
+          "requirements": "Compares FBA and pFBA growth/secretion regimes and explains where they agree or differ. Score 100% if FBA and pFBA growth rates are numerically compared and total flux norms are reported for both; 67% if only growth rates are compared without flux-norm comparison; 33% if pFBA was run but not compared to FBA; 0% otherwise.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b02-result-figure",
+          "requirements": "Produces at least one labelled heatmap or contour plot for growth and acetate secretion with units.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b02-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, characterises the acetate-overflow regime boundary with quantitative support, and gives a one-paragraph mechanistic interpretation of why E. coli shifts from respiratory to acetate-secreting overflow metabolism at the identified nutrient threshold.",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b02-repro",
+      "requirements": "The model ID, medium, grid values, solver, and thresholds are recorded for rerun reproducibility.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b02-repro-model",
+          "requirements": "The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object used.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b02-repro-config",
+          "requirements": "Saves model ID, objective reaction, nutrient grid, and secretion threshold in config or results metadata.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b02-repro-solver",
+          "requirements": "Records solver backend, tolerance, COBRApy version, and rerun consistency for the aerobic reference point.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/rubrics/B03.json

@@ -0,0 +1,166 @@
+{
+  "id": "B03",
+  "requirements": "A credible anaerobic E. coli lactate-overproduction study using COBRApy that validates anaerobic growth, computes a lactate production envelope, screens a focused central-carbon/fermentation knockout set, and explains top candidates mechanistically.",
+  "judging_note": "Accept iJO1366, iAF1260, or comparable curated E. coli models. Score coherent anaerobic setup and mechanistic KO interpretation over exact gene names.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b03-code",
+      "requirements": "COBRApy code implements anaerobic medium, lactate objective/envelope, and knockout screening.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b03-code-medium",
+          "requirements": "Loads an E. coli BIGG model, sets glucose uptake, closes oxygen uptake, and names biomass and lactate exchange reactions.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b03-code-envelope",
+          "requirements": "Computes biomass-vs-lactate production envelope by constraining biomass fractions and optimizing or FVA-bounding lactate secretion.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b03-code-ko",
+          "requirements": "Screens a focused set of no more than 100 central-carbon and fermentation genes with model state restored between knockouts.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b03-code-yield",
+          "requirements": "Computes lactate yield as lactate secretion divided by absolute glucose uptake, separate from flux.",
+          "weight": 4,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b03-exec",
+      "requirements": "Execution produces valid anaerobic WT, envelope, and KO tables.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b03-exec-validity",
+          "requirements": "WT anaerobic FBA is optimal with positive biomass and at least 85% of KO simulations solve or are explicitly labelled infeasible/lethal.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b03-exec-physics",
+          "requirements": "Mass-balance check passes for the loaded anaerobic model; no aerobic by-products (O2 uptake or CO2 via respiration) appear under strictly anaerobic conditions without explanation; biomass flux is positive under anaerobic glucose reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b03-exec-artifacts",
+          "requirements": "Writes results.json plus machine-readable envelope and KO-ranking tables.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b03-results",
+      "requirements": "Results quantify lactate production, growth tradeoffs, and KO mechanisms.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b03-result-h1",
+          "requirements": "Reports anaerobic WT growth and confirms oxygen uptake is zero or bounded closed. Score 100% if positive anaerobic WT biomass flux is reported with O2 exchange confirmed ≤0 mmol/gDW/h; 67% if growth is positive but O2 status is not explicitly confirmed; 33% if anaerobic feasibility is stated without a numeric growth value; 0% otherwise.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b03-result-h2",
+          "requirements": "Shows lactate secretion changes across biomass fractions and identifies whether production is growth-coupled, competing, or partially coupled. Score 100% if lactate secretion flux is non-decreasing as biomass fraction decreases across ≥8 of 10 envelope grid points; 67% if non-decreasing over ≥6 of 10 points; 33% if ≥4 of 10 points; 0% otherwise.",
+          "weight": 10,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b03-result-h3",
+          "requirements": "Top 5 KO list includes growth fraction, lactate flux/yield, and pathway-level rationale for at least top 3 targets. Score 100% if ≥3 of the top-5 KOs disrupt canonical competing by-product pathways (acetate, ethanol, formate, or succinate routes) or include a specific mechanistic explanation; 67% if 2 of 5 meet this criterion; 33% if 1 of 5; 0% otherwise.",
+          "weight": 12,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b03-result-figure",
+          "requirements": "Produces a labelled lactate envelope or KO ranking figure with units.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b03-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, names the top lactate-improving knockouts with their gene/enzyme identity, and gives a one-paragraph mechanistic interpretation (which competing pathways the KOs disable, why this redirects anaerobic flux to lactate).",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b03-repro",
+      "requirements": "Medium, model, gene set, solver, and thresholds are reproducibly recorded.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b03-repro-model",
+          "requirements": "The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce the anaerobic medium setup.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b03-repro-runcard",
+          "requirements": "Saves anaerobic medium bounds, biomass threshold, KO gene set, and lactate reaction ID.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b03-repro-solver",
+          "requirements": "Records COBRApy/solver versions and confirms rerun consistency for WT growth.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/rubrics/B04.json

@@ -0,0 +1,166 @@
+{
+  "id": "B04",
+  "requirements": "A credible S. cerevisiae ethanol-yield study that loads a yeast GSMM, validates growth, sweeps oxygen uptake, swaps carbon sources, and reports ethanol secretion/yield with interpretable figures.",
+  "judging_note": "The main challenge is adapting the pipeline beyond E. coli while remaining within COBRApy/BIGG capabilities. Accept iMM904 or another documented yeast GSMM.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b04-code",
+      "requirements": "The code implements yeast model loading, medium setup, oxygen sweep, carbon-source swap, and ethanol-yield calculations.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b04-code-model",
+          "requirements": "Loads iMM904 or comparable yeast model, identifies biomass, glucose, oxygen, and ethanol exchange reactions, and validates positive reference growth.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b04-code-o2",
+          "requirements": "Sweeps oxygen uptake across at least 8 values while keeping substrate uptake defined and restoring state between runs.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b04-code-carbon",
+          "requirements": "Compares at least 4 carbon sources where model reactions exist and explicitly records unsupported sources.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b04-code-yield",
+          "requirements": "Calculates ethanol yield using absolute substrate uptake and separates yield from raw secretion flux.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b04-exec",
+      "requirements": "Execution produces oxygen-sweep and carbon-source artifacts without crashing.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b04-exec-status",
+          "requirements": "Reference glucose condition is optimal and at least 80% of supported oxygen/carbon conditions solve or are clearly marked infeasible.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b04-exec-physics",
+          "requirements": "Mass-balance check passes for the yeast GSMM; no non-physical fluxes appear under aerobic glucose reference (e.g. ethanol secretion should be low under full aerobic conditions); biomass flux is positive under yeast glucose-minimal reference. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b04-exec-artifacts",
+          "requirements": "Writes results.json, oxygen_sweep.csv, and carbon_source_comparison.csv or equivalent structured outputs.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b04-results",
+      "requirements": "Results quantify oxygen and substrate effects on ethanol secretion/yield.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b04-result-h1",
+          "requirements": "Reports yeast reference growth with model ID and medium definition. Score 100% if positive yeast biomass flux and ethanol secretion are both reported under aerobic glucose-minimal medium with optimal solver status and a named model ID; 67% if growth is positive but ethanol is not reported; 33% if feasibility is stated without numeric values; 0% otherwise.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b04-result-h2",
+          "requirements": "Quantifies how ethanol flux or yield changes as oxygen decreases and identifies the low-oxygen fermentation regime. Score 100% if ethanol flux or yield is non-decreasing as oxygen uptake decreases across ≥6 of 8 oxygen-sweep points; 67% if non-decreasing over ≥4 of 8 points; 33% if the trend is described without a quantitative grid; 0% otherwise.",
+          "weight": 12,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b04-result-h3",
+          "requirements": "Compares growth and ethanol yield across carbon sources with a biological interpretation. Score 100% if at least 3 carbon sources are compared with distinct numeric ethanol yields and a biological interpretation; 67% if 2 sources are compared with numeric differences; 33% if the comparison is qualitative or only one carbon source produces positive growth; 0% otherwise.",
+          "weight": 10,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b04-result-figure",
+          "requirements": "Produces labelled figures for oxygen sweep and/or carbon-source yield comparison.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b04-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, names the oxygen regime and carbon source that maximise ethanol yield, and gives a one-paragraph mechanistic interpretation of why oxygen restriction switches yeast from respiratory growth to fermentative ethanol production.",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b04-repro",
+      "requirements": "Model source, medium, oxygen grid, carbon-source reactions, solver, and versions are documented.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b04-repro-model",
+          "requirements": "The loaded yeast GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID, version, and source URL in the writeup; a downstream user can locate the exact model object used.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b04-repro-config",
+          "requirements": "Saves reaction IDs, bounds, grid values, and carbon-source list in config or results metadata.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b04-repro-version",
+          "requirements": "Records COBRApy, solver, model source/version, and rerun consistency for reference growth.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/rubrics/B05.json

@@ -0,0 +1,166 @@
+{
+  "id": "B05",
+  "requirements": "A credible M. tuberculosis essentiality study using COBRApy that validates a pathogen GSMM, runs single-gene and/or single-reaction deletions, identifies essential targets, and prioritises interpretable drug-target hypotheses.",
+  "judging_note": "Accept gene deletions, reaction deletions, or both depending on model GPR support. Penalize unsupported biological claims more than absence of external drug databases.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b05-code",
+      "requirements": "The code loads the pathogen model, validates medium, runs deletion screens, and annotates essential targets.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b05-code-model",
+          "requirements": "Loads iNJ661 or a documented M. tuberculosis GSMM, defines medium/objective, and validates positive WT biomass.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b05-code-deletion",
+          "requirements": "Runs single-gene deletion when GPR rules are usable, or single-reaction deletion as an explicit fallback/complement.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b05-code-threshold",
+          "requirements": "Defines essentiality threshold such as growth <5% or <10% of WT and applies it consistently.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b05-code-annotation",
+          "requirements": "Extracts subsystem, reaction name, gene name, GPR, or other model-native annotations for top targets.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b05-exec",
+      "requirements": "Deletion runs complete with interpretable status handling and structured outputs.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b05-exec-status",
+          "requirements": "WT reference condition is optimal and at least 85% of deletion simulations solve or are explicitly marked infeasible/lethal.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b05-exec-physics",
+          "requirements": "Mass-balance check passes for the loaded pathogen GSMM; WT biomass is positive under the documented medium; deletion table contains no negative growth values; model GPR rules produce logically consistent essential gene predictions. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b05-exec-artifacts",
+          "requirements": "Writes results.json plus deletion table with growth, growth_fraction, status, and essential flag.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b05-results",
+      "requirements": "Results identify essential targets and provide conservative biological interpretation.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b05-result-h1",
+          "requirements": "Reports WT growth, medium, model ID, and validation caveats. Score 100% if positive WT biomass flux is reported under the documented medium with optimal solver status and model ID named; 67% if growth is positive but medium is not explicitly documented; 33% if feasibility is stated without a numeric value; 0% otherwise.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b05-result-h2",
+          "requirements": "Reports essential target count and distribution of growth fractions after deletion. Score 100% if ≥5 essential genes/reactions are identified with growth fractions below the stated threshold and the deletion table covers ≥50 candidates; 67% if essential targets are identified but the candidate set covers fewer than 50 entries; 33% if essential targets are named without a deletion table; 0% otherwise.",
+          "weight": 10,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b05-result-h3",
+          "requirements": "Top 10 targets include subsystem/pathway and one-sentence rationale; claims are framed as modelling hypotheses, not validated drugs. Score 100% if ≥3 of the top-10 prioritised targets belong to named core metabolic subsystems (cell-wall precursor, cofactor, lipid, energy, or amino-acid biosynthesis) with at least a one-sentence pathway rationale; 67% if 2 of 10 with rationale; 33% if 1 of 10 with rationale; 0% otherwise.",
+          "weight": 12,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b05-result-figure",
+          "requirements": "Produces a labelled essentiality histogram, target ranking plot, or subsystem enrichment chart.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b05-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, names the top prioritised essential targets with subsystem/enzyme identity, and gives a one-paragraph mechanistic rationale for why each target's disruption is lethal. Claims are explicitly framed as computational hypotheses requiring wet-lab validation.",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b05-repro",
+      "requirements": "Model, medium, deletion type, threshold, solver, and annotations are reproducibly recorded.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b05-repro-model",
+          "requirements": "The loaded M. tuberculosis GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID, version, or local file hash in the writeup; a downstream user can locate the exact model object used.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b05-repro-config",
+          "requirements": "Saves model source, medium bounds, objective, deletion mode, and essentiality threshold.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b05-repro-version",
+          "requirements": "Records COBRApy/solver versions and confirms rerun consistency for WT growth and essential count.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/rubrics/B06.json

@@ -0,0 +1,166 @@
+{
+  "id": "B06",
+  "requirements": "A credible E. coli carbon-source robustness study that loads a validated GSMM, swaps carbon sources, reports growth/secretion profiles, and identifies condition-dependent essential genes from a focused gene set.",
+  "judging_note": "Score explicit medium handling and condition-dependent interpretation. It is acceptable if some carbon sources are unsupported, provided they are recorded and not silently dropped.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b06-code",
+      "requirements": "The code implements carbon-source medium swaps, FBA/pFBA, and focused essentiality comparison.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b06-code-model",
+          "requirements": "Loads E. coli BIGG model, identifies biomass and candidate carbon exchange reactions, and defines a minimal medium policy.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b06-code-swap",
+          "requirements": "Closes background carbon uptake and opens one carbon source at a time with documented uptake bounds.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Experimental Setup"
+        },
+        {
+          "id": "b06-code-secretion",
+          "requirements": "Runs FBA/pFBA per source and records major positive exchange fluxes.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b06-code-essentiality",
+          "requirements": "Runs focused single-gene deletions for feasible carbon sources and computes growth fractions relative to each source-specific WT.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b06-exec",
+      "requirements": "Execution produces carbon-source and essentiality artifacts.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b06-exec-validity",
+          "requirements": "Glucose reference is optimal and at least two non-glucose sources are either feasible or clearly marked unsupported/infeasible.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b06-exec-physics",
+          "requirements": "Mass-balance check passes for the E. coli GSMM under each tested carbon source; no carbon source produces negative biomass or non-physical secretion without explicit handling; model state is restored between carbon-source conditions. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b06-exec-artifacts",
+          "requirements": "Writes results.json, carbon-source table, and gene-by-condition growth-fraction matrix.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b06-results",
+      "requirements": "Results compare carbon sources and identify condition-dependent vulnerabilities.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b06-result-h1",
+          "requirements": "Reports feasible carbon-source count and growth rates, including unsupported-source handling. Score 100% if ≥3 carbon sources including glucose support positive biomass growth with explicit exchange bounds documented; 67% if 2 sources support positive growth; 33% if only glucose is feasible with explicit confirmation that other sources are infeasible or unsupported; 0% otherwise.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b06-result-h2",
+          "requirements": "Compares secretion profiles and growth rates across sources with quantitative differences. Score 100% if growth rates differ by >10% between at least 2 feasible carbon sources AND major secretion fluxes differ qualitatively across sources; 67% if growth rate differences are reported without secretion-profile comparison; 33% if differences are noted qualitatively without quantification; 0% otherwise.",
+          "weight": 9,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b06-result-h3",
+          "requirements": "Identifies genes with large growth-fraction changes across carbon sources and explains pathway context for top hits. Score 100% if ≥3 genes with condition-dependent essentiality changes (essential under one carbon source but not another) are identified with pathway context; 67% if 2 genes are identified; 33% if 1 gene is identified with pathway context; 0% otherwise.",
+          "weight": 12,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b06-result-figure",
+          "requirements": "Produces labelled bar plots and/or heatmaps for growth, secretion, and condition-dependent essentiality.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b06-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, names supported and unsupported carbon sources, identifies condition-dependent essential genes, and gives a one-paragraph mechanistic interpretation linking carbon-source specific metabolism to the observed vulnerability changes.",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b06-repro",
+      "requirements": "Carbon-source definitions, gene set, thresholds, solver, and model are documented.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b06-repro-model",
+          "requirements": "The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce each carbon-source medium setup.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b06-repro-runcard",
+          "requirements": "Saves carbon exchange IDs, uptake bounds, oxygen setting, gene list, and essentiality threshold.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b06-repro-version",
+          "requirements": "Records COBRApy/solver versions and rerun consistency for glucose and at least one non-glucose source.",
+          "weight": 5,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/rubrics/B07.json

@@ -0,0 +1,166 @@
+{
+  "id": "B07",
+  "requirements": "A reproducible FBA-method benchmark on E. coli that compares standard FBA, pFBA, loopless FBA when available, and FVA near optimum using consistent medium, metrics, runtime logging, and interpretable figures.",
+  "judging_note": "This is a protocol-quality task. Reward robust implementation, structured outputs, and careful handling of unavailable loopless solvers over biological novelty.",
+  "weight": 1,
+  "sub_tasks": [
+    {
+      "id": "b07-code",
+      "requirements": "The code implements method comparison with consistent model state and metrics.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b07-code-model",
+          "requirements": "Loads E. coli BIGG model, sets medium and biomass objective, and validates reference FBA.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b07-code-methods",
+          "requirements": "Runs standard FBA, pFBA, and loopless FBA or records a justified loopless-unavailable status.",
+          "weight": 7,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b07-code-fva",
+          "requirements": "Runs FVA at a documented fraction of optimum and computes interval widths and rigid/flexible labels.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Method Implementation"
+        },
+        {
+          "id": "b07-code-metrics",
+          "requirements": "Computes total absolute flux, active reaction count, objective value, solver status, and runtime for each method.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b07-exec",
+      "requirements": "Execution produces complete benchmark artifacts and handles optional method failures explicitly.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b07-exec-status",
+          "requirements": "Standard FBA and pFBA solve optimally; loopless and FVA either solve or have explicit, non-crashing failure records.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b07-exec-physics",
+          "requirements": "Mass-balance check passes for the loaded model under the benchmark medium; FBA, pFBA, and loopless FBA all return non-negative biomass fluxes; no negative flux norms or ill-conditioned solver status appear without explicit logging. This is the biology-validity gate equivalent to physics-validity checks in the physics rubric.",
+          "weight": 7.0,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b07-exec-artifacts",
+          "requirements": "Writes results.json, method comparison table, FVA table, and method flux tables.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Code Execution",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b07-results",
+      "requirements": "Results compare objective preservation, parsimony, and reaction variability.",
+      "weight": 3,
+      "sub_tasks": [
+        {
+          "id": "b07-result-h1",
+          "requirements": "Reports objective differences among methods and checks agreement within solver tolerance where methods succeed. Score 100% if FBA and pFBA biomass objectives agree within 1% relative error and any available loopless FBA also agrees within 1%; 67% if FBA and pFBA agree but loopless comparison is missing or failed with explicit explanation; 33% if only qualitative agreement is stated; 0% otherwise.",
+          "weight": 9,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Evaluation, Metrics & Benchmarking"
+        },
+        {
+          "id": "b07-result-h2",
+          "requirements": "Quantifies pFBA parsimony using total flux norm and active reaction count compared with standard FBA. Score 100% if pFBA total absolute flux norm is numerically lower than FBA AND active reaction count is lower under pFBA; 67% if flux norm is lower but active reaction count is not compared; 33% if pFBA parsimony is stated qualitatively without numeric comparison; 0% otherwise.",
+          "weight": 10,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b07-result-h3",
+          "requirements": "Summarizes FVA width distribution and names central-carbon reactions that are rigid or flexible near optimum. Score 100% if FVA interval widths are reported for all model reactions and ≥5 central-carbon reactions are classified as rigid (interval width < 1 mmol/gDW/h) with subsystem identification; 67% if rigid/flexible classification is reported without central-carbon specifics; 33% if FVA results are reported without classification; 0% otherwise.",
+          "weight": 10,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b07-result-figure",
+          "requirements": "Produces labelled method-comparison and/or FVA-width figures with units and clear legends.",
+          "weight": 8,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        },
+        {
+          "id": "b07-result-writeup",
+          "requirements": "The README or results writeup discusses each hypothesis outcome, summarises the method-level comparison (parsimony, variability, runtime), and gives a one-paragraph protocol recommendation stating which method is most suitable for reproducible phenotype prediction and why.",
+          "weight": 6.0,
+          "sub_tasks": [],
+          "task_category": "Result Analysis",
+          "finegrained_task_category": "Logging, Analysis & Presentation"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    },
+    {
+      "id": "b07-repro",
+      "requirements": "The benchmark is reproducible with pinned model, medium, tolerance, solver, and code path.",
+      "weight": 2,
+      "sub_tasks": [
+        {
+          "id": "b07-repro-model",
+          "requirements": "The loaded E. coli GSMM is either persisted (models/<model_id>.json) or pinned by BIGG model ID and source URL in the writeup; a downstream user can locate the exact model object and reproduce the benchmark medium.",
+          "weight": 5.0,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b07-repro-config",
+          "requirements": "Saves model ID, objective, medium bounds, FVA fraction, active-flux tolerance, and method list.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        },
+        {
+          "id": "b07-repro-version",
+          "requirements": "Records COBRApy, optlang, solver backend/version, and runtime environment.",
+          "weight": 6,
+          "sub_tasks": [],
+          "task_category": "Code Development",
+          "finegrained_task_category": "Reproducibility"
+        }
+      ],
+      "task_category": null,
+      "finegrained_task_category": null
+    }
+  ],
+  "task_category": null,
+  "finegrained_task_category": null
+}

tasks/biology/topics.yaml

@@ -0,0 +1,99 @@
+# =============================================================================
+# ARC-Bench Biology — constraint-based metabolic modelling topics
+# -----------------------------------------------------------------------------
+# All biology topics drive the Biology-Agent integration:
+#   experiment.mode = biology_agent  →  BiologyAgentSandbox  →  Claude Code
+#   with /home/shiqiu/Biology-Agent skills + agents installed into ~/.claude/.
+# =============================================================================
+
+topics:
+  - id: B01
+    topic: "E. coli succinate-production strain optimization via constraint-based knockout screen on iAF1260"
+    domains: ["systems-biology", "metabolic-engineering", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iAF1260"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B02
+    topic: "E. coli acetate overflow metabolism phase map under glucose and oxygen limitation"
+    domains: ["systems-biology", "constraint-based-modelling", "phase-plane-analysis"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iJO1366"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B03
+    topic: "Anaerobic E. coli lactate-overproduction knockout screen"
+    domains: ["systems-biology", "metabolic-engineering", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iJO1366"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B04
+    topic: "S. cerevisiae ethanol yield under oxygen limitation and carbon-source changes"
+    domains: ["systems-biology", "metabolic-engineering", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iMM904"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B05
+    topic: "M. tuberculosis condition-specific essentiality and metabolic drug-target prioritisation"
+    domains: ["systems-biology", "drug-target-prioritisation", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iNJ661"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B06
+    topic: "E. coli carbon-source robustness and condition-dependent essential genes"
+    domains: ["systems-biology", "essentiality-analysis", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iJO1366"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"
+
+  - id: B07
+    topic: "Reproducible FBA protocol benchmark: FBA vs pFBA vs loopless/FVA on E. coli"
+    domains: ["systems-biology", "method-benchmarking", "constraint-based-modelling"]
+    metric_key: "primary_metric"
+    metric_direction: "maximize"
+    bigg_model: "iJO1366"
+    source_skills:
+      - "Biology-Agent/skills/gsmm-builder"
+      - "Biology-Agent/skills/gsmm-validator"
+      - "Biology-Agent/skills/fba-simulator"
+      - "Biology-Agent/skills/flux-analyzer"
+      - "Biology-Agent/skills/mfa-pipeline-orchestrator"

tasks/meta_paper_quality.json

@@ -0,0 +1,211 @@
+{
+    "id": "meta-paper-quality-root",
+    "requirements": "Paper-quality meta-rubric applied to ALL ARC-Bench topics (ML / P / B). Grades the final deliverables — paper_final.md, charts/*, code/*, submission/README.md — for paper writing quality, code orchestration, visual layout, and content accuracy. This rubric is SEPARATE from the per-topic science rubric and is graded MANUALLY by a Claude Code agent invoked via experiments/arc_bench/scripts/judge_paper_manual.sh. The bench pipeline does NOT auto-run this gate. The combined overall score = (per-topic-science × 100 + meta-paper-quality × 86) / (100 + 86) ≈ 54% science + 46% meta.",
+    "judging_note": "Score on substance, not surface polish. A 750-word abstract with shallow content is worse than a 200-word abstract with a clear claim+method+result. Penalize fabrication (numbers in writeup not in artifacts) heavily, hedge claims to match evidence, and credit modular code even if it could be shorter. For figures, the test is communicative function: can a reader understand the message from the figure alone? When ambiguous, assign partial credit (0.33 / 0.5 / 0.67) rather than forcing binary 0/1.",
+    "weight": 1,
+    "scoring_target": "deliverables",
+    "input_artifacts": [
+        "stage-22/paper_final.md",
+        "stage-19/paper_revised.md",
+        "stage-17/paper_draft.md",
+        "deliverables/paper_final.md",
+        "deliverables/charts/",
+        "deliverables/code/",
+        "submission/README.md",
+        "submission/code/",
+        "stage-14/charts/",
+        "stage-14/experiment_summary.json"
+    ],
+    "sub_tasks": [
+        {
+            "id": "meta-paper-content",
+            "requirements": "Paper writing quality across the IMRAD structure (Introduction / Methods / Results / Discussion). Grades clarity, structure, faithfulness to results, and academic conventions — NOT surface polish or word count.",
+            "weight": 5,
+            "sub_tasks": [
+                {
+                    "id": "mpq-abstract",
+                    "requirements": "Abstract captures the four required moves: (1) what problem/question, (2) what method/approach was taken, (3) what was found numerically, (4) what the contribution is. Length 150-300 words. NO numbers should appear in the abstract that don't appear in the results section. Subscore 1.0 if all four moves present and grounded; 0.67 if 3 of 4; 0.33 if 2 of 4; 0 otherwise.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - Abstract"
+                },
+                {
+                    "id": "mpq-introduction",
+                    "requirements": "Introduction situates the work: prior approaches, the specific gap addressed, and the contribution. Avoids 'in this paper we' boilerplate without substance. Cites real prior work (≥3 citations) relevant to the topic. Subscore on coherence of the motivation chain — can a reader who knows the field tell what's new vs prior art?",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - Introduction"
+                },
+                {
+                    "id": "mpq-method-clarity",
+                    "requirements": "Methods section enables independent reproduction by someone with domain knowledge but no prior context. Names: (a) datasets / models used with versions or sources, (b) algorithms / settings / hyperparameters in numbers, (c) evaluation protocol (seeds, splits, metrics). Acceptable: equations, pseudocode, or precise prose. Unacceptable: 'we used standard techniques' without specifics.",
+                    "weight": 6,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - Methods"
+                },
+                {
+                    "id": "mpq-results-grounded",
+                    "requirements": "Every numeric claim in the Results section is traceable to a value in the artifacts (experiment_summary.json, results.json, CSV outputs). No invented numbers. Comparisons (X > Y, A improves over B) are backed by stated differences and uncertainty. Tables and figures are referenced from the prose. Subscore: count the numeric claims, mark how many are traceable; subscore = traceable / total.",
+                    "weight": 8,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - Results"
+                },
+                {
+                    "id": "mpq-discussion",
+                    "requirements": "Discussion section addresses: (a) what the results imply / mechanistic interpretation, (b) at least one threat to validity or limitation honestly stated (small N, seed variance, dataset bias, methodological caveat), (c) suggested follow-up work. Penalize 'no limitations' or generic 'future work' boilerplate.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - Discussion"
+                },
+                {
+                    "id": "mpq-citations",
+                    "requirements": "Citations are valid (real DOIs / arXiv IDs / venue names) and formatted consistently. Verify against references.bib: every cited key resolves, no hallucinated authors or titles. Sample 3-5 citations and check existence (a quick web lookup or arXiv ID validation). Bibliography format is internally consistent (all entries follow the same style).",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Paper Quality",
+                    "finegrained_task_category": "Writing - References"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "meta-code-orchestration",
+            "requirements": "Code orchestration quality of the experiment package. Looks at stage-22/code/ or submission/code/ — how the actual experimental code is organized, documented, and made reusable. Distinct from the science rubric's code leaves (which check whether the method is implemented at all).",
+            "weight": 3,
+            "sub_tasks": [
+                {
+                    "id": "mco-modular",
+                    "requirements": "Code is split into logical modules with clear responsibilities (data loading, model definition, training/inference, evaluation, plotting), NOT a single monolithic main.py. Each module is <500 lines. Imports are explicit (no `from X import *`). Subscore: 1.0 if cleanly split into ≥3 named modules; 0.67 if 2 modules; 0.33 if monolithic but reasonable internal structure; 0 if spaghetti.",
+                    "weight": 5,
+                    "sub_tasks": [],
+                    "task_category": "Code Orchestration",
+                    "finegrained_task_category": "Code Quality - Modularity"
+                },
+                {
+                    "id": "mco-reproducible",
+                    "requirements": "A reader could run the experiment from the code package alone. Means: (a) a README or main.py docstring explains the entry point, (b) dependencies are listed (requirements.txt, setup.py, or imports clearly inferable), (c) random seeds are settable / documented, (d) data sources are specified (sklearn datasets, BIGG model id, etc.). NO hardcoded absolute paths to the original author's machine.",
+                    "weight": 6,
+                    "sub_tasks": [],
+                    "task_category": "Code Orchestration",
+                    "finegrained_task_category": "Code Quality - Reproducibility"
+                },
+                {
+                    "id": "mco-readable",
+                    "requirements": "Identifier names communicate intent (no `x1`, `tmp`, `foo`). Function/class docstrings explain purpose for non-trivial code. Public API has type hints for at least the function signatures. Limited inline comments (only WHY when non-obvious), NOT line-by-line narration. Mix of these signals → 0.0 to 1.0.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Code Orchestration",
+                    "finegrained_task_category": "Code Quality - Readability"
+                },
+                {
+                    "id": "mco-no-dead-code",
+                    "requirements": "No commented-out code blocks (more than 3 contiguous commented lines). No unused imports. No functions/classes defined but never called. No TODO/FIXME markers without ownership. Penalize 'kitchen-sink' files that import and define dozens of things to use only a few.",
+                    "weight": 3,
+                    "sub_tasks": [],
+                    "task_category": "Code Orchestration",
+                    "finegrained_task_category": "Code Quality - Hygiene"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "meta-visual-layout",
+            "requirements": "Figure quality of the deliverable charts (stage-22/deliverables/charts/, submission/code/*.png|*.pdf). Grades whether each figure communicates its message effectively to a reader who has not read the prose.",
+            "weight": 3,
+            "sub_tasks": [
+                {
+                    "id": "mvl-axes-labeled",
+                    "requirements": "Every figure has both axes labeled with quantity + unit (e.g., 'Growth rate (1/h)', 'Cross section (pb)', 'Accuracy (%)'). Tick labels are readable at the figure's printed size. Subscore: fraction of figures (in the deliverable charts dir) that pass.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Visual Layout",
+                    "finegrained_task_category": "Figure - Axes"
+                },
+                {
+                    "id": "mvl-legend-present",
+                    "requirements": "Multi-series figures (≥2 lines, bars, or categories) have a legend that identifies each series. Single-series figures do NOT need a legend. Legend placement does not occlude data. Color choices distinguishable by colorblind readers (avoid pure red+green together without a marker shape difference).",
+                    "weight": 3,
+                    "sub_tasks": [],
+                    "task_category": "Visual Layout",
+                    "finegrained_task_category": "Figure - Legend"
+                },
+                {
+                    "id": "mvl-caption-quality",
+                    "requirements": "Each figure has a caption that (a) names what is plotted, (b) names the dataset / condition / model used, (c) states the key takeaway in one sentence. A caption that just says 'Figure 1: results' is insufficient. Captions are in the paper text near the figure reference.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Visual Layout",
+                    "finegrained_task_category": "Figure - Caption"
+                },
+                {
+                    "id": "mvl-figure-relevance",
+                    "requirements": "Each figure earns its place — it shows information that supports a claim in the paper. No decorative figures, no figures duplicated across stages (the same chart appearing twice from different stage outputs), no figures showing trivially obvious results without value. Subscore on the fraction of figures that pass this 'earned place' test.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Visual Layout",
+                    "finegrained_task_category": "Figure - Relevance"
+                },
+                {
+                    "id": "mvl-color-accessibility",
+                    "requirements": "Color palette is colorblind-safe (avoid red+green only); important distinctions are reinforced by shape, line style, or hatching in addition to color. Background does not interfere with foreground (no dark text on dark background). Resolution is at least 150 DPI for raster figures.",
+                    "weight": 2,
+                    "sub_tasks": [],
+                    "task_category": "Visual Layout",
+                    "finegrained_task_category": "Figure - Accessibility"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        },
+        {
+            "id": "meta-content-accuracy",
+            "requirements": "Faithfulness of the writeup to the underlying artifacts. The single biggest fabrication risk in autonomous paper generation is invented numbers; this bucket carries the largest weight.",
+            "weight": 4,
+            "sub_tasks": [
+                {
+                    "id": "mca-no-fabrication",
+                    "requirements": "STRICT: every numeric value in the paper (abstract, methods, results, tables) must trace to a value in the artifacts (results.json, experiment_summary.json, CSV outputs, figure data). Sample 5-10 numbers across the paper; mark each as traceable / not-traceable; subscore = traceable_count / sampled_count. Any single invented number for a key metric (e.g. final accuracy, final growth rate) caps the leaf at 0.5 regardless of other traceable numbers.",
+                    "weight": 8,
+                    "sub_tasks": [],
+                    "task_category": "Content Accuracy",
+                    "finegrained_task_category": "Faithfulness - Numbers"
+                },
+                {
+                    "id": "mca-correct-units",
+                    "requirements": "Units are consistent and physically correct throughout. ML: percentages stay as percent vs fractions consistently; runtime in seconds vs minutes consistently. HEP: GeV vs MeV not mixed, cross sections in pb vs fb consistent. Biology: mmol/gDW/h is the standard; do not mix with mM. Penalize 'unit-less' numbers where a unit is needed.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Content Accuracy",
+                    "finegrained_task_category": "Faithfulness - Units"
+                },
+                {
+                    "id": "mca-claim-strength",
+                    "requirements": "Strength of claims matches strength of evidence. 'X significantly outperforms Y' requires a statistical test result; 'X tends to be better than Y' requires a directional difference; 'X may be useful' requires only intuition. Penalize uncalibrated 'state of the art', 'novel', 'first to show' without specific support. Hedging language ('appears', 'suggests', 'is consistent with') should appear when the evidence is suggestive.",
+                    "weight": 5,
+                    "sub_tasks": [],
+                    "task_category": "Content Accuracy",
+                    "finegrained_task_category": "Faithfulness - Claim Calibration"
+                },
+                {
+                    "id": "mca-internal-consistency",
+                    "requirements": "The same number / claim appears identically across the paper. Abstract says 'accuracy 92.3%'; Results says 'accuracy 92.3%' (not 92.5%); Conclusion says 'achieved 92.3%'. Tables match prose. Figure captions match figure content. Penalize ANY internal contradiction even if both versions might be 'plausible'.",
+                    "weight": 4,
+                    "sub_tasks": [],
+                    "task_category": "Content Accuracy",
+                    "finegrained_task_category": "Faithfulness - Internal Consistency"
+                }
+            ],
+            "task_category": null,
+            "finegrained_task_category": null
+        }
+    ],
+    "task_category": null,
+    "finegrained_task_category": null
+}

tasks/ml/manifests/ML01.yaml

@@ -0,0 +1,89 @@
+# ============================================================================
+# T01 — Dropout regularization strategies for shallow MLPs
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML01
+title: "Dropout regularization strategies on shallow tabular MLPs"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Dropout-style regularization is a cornerstone of neural-net training for
+  overfitting control, but its impact on tabular classification with shallow
+  MLPs remains subtle. Standard element-wise dropout is the default; spatial
+  (feature-wise) dropout treats correlated input features as a block; MC /
+  variational dropout keeps dropout active at test time and averages over
+  samples. For tabular data with small feature counts and mild non-linearity,
+  these variants can produce equal test-set accuracy yet very different
+  probability calibration — a known gotcha that makes accuracy-only
+  evaluation misleading.
+
+  A credible CPU-scale study of this topic (a) compares at least three
+  dropout variants against a no-dropout control on multiple sklearn
+  classification benchmarks, (b) reports both accuracy and calibration
+  metrics (ECE, NLL, Brier), (c) averages over ≥5 seeds with statistical
+  tests, and (d) checks for ablation integrity (that different dropout
+  variants actually produce different output distributions, not just the
+  same network). The research question is: *does the choice of dropout
+  variant matter more for calibration than for accuracy on tabular data?*
+
+hypotheses:
+  - id: H1
+    statement: "Variational / MC dropout produces lower Expected Calibration Error (ECE) than standard element-wise dropout on at least 2 of 3 tabular classification benchmarks, averaged over ≥5 seeds."
+    measurable: true
+  - id: H2
+    statement: "Test-set accuracy differs by <2 absolute percentage points between dropout variants on datasets where all methods exceed 95% accuracy — i.e. accuracy is non-discriminative and calibration is needed to distinguish methods."
+    measurable: true
+  - id: H3
+    statement: "The no-dropout control is strictly worse than at least one dropout variant in ECE on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Does the choice of dropout variant (standard / spatial / variational) matter more for probability calibration than for test accuracy on tabular classification with shallow MLPs?"
+  conditions:
+    - name: "no_dropout"
+      description: "Shallow MLP (2 hidden layers, 64 units, ReLU) trained with no dropout. Control condition."
+    - name: "standard_dropout_p30"
+      description: "Same MLP with standard element-wise dropout at p=0.3 between hidden layers."
+    - name: "standard_dropout_p50"
+      description: "Same MLP with standard element-wise dropout at p=0.5."
+    - name: "spatial_dropout_p30"
+      description: "Same MLP with feature-blockwise dropout at p=0.3 applied to the input layer."
+    - name: "mc_dropout_p30_T20"
+      description: "Same MLP with standard p=0.3 dropout left active at test time; prediction is the mean of T=20 stochastic forward passes."
+  baselines:
+    - "no_dropout is the no-regularization baseline"
+  metrics:
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Fraction correctly classified on held-out 20% test split, mean over 5 seeds."
+    - name: "ece"
+      direction: "minimize"
+      description: "Expected Calibration Error (15 bins) on the test split, mean over 5 seeds."
+    - name: "nll"
+      direction: "minimize"
+      description: "Mean per-example negative log-likelihood on the test split."
+    - name: "brier"
+      direction: "minimize"
+      description: "Mean Brier score on the test split."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 240
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML01.json"

tasks/ml/manifests/ML02.yaml

@@ -0,0 +1,98 @@
+# ============================================================================
+# T02 — Ensemble-method comparison for non-linear regression under noise
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML02
+title: "Bagging vs boosting vs stacking for noisy non-linear regression"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Ensemble learning is often presented as a broadly reliable way to improve
+  predictive performance, but different ensemble families can react very
+  differently to observation noise in regression. Bagging primarily reduces
+  variance by averaging many weakly correlated models. Boosting is sequential
+  and can aggressively fit residuals, which can help on structured non-linear
+  signals but may overfit high-noise targets if regularization is not tuned.
+  Stacking combines heterogeneous base learners through a meta-model and may
+  benefit from diversity, but it also introduces extra fitting stages that can
+  be sensitive to data size and leakage handling.
+
+  A focused CPU-scale benchmark can probe these tradeoffs using synthetic and
+  lightweight sklearn regression datasets where non-linearity and noise are
+  controllable. By varying noise level explicitly in generated data, the study
+  can test whether relative ranking among bagging, boosting, and stacking is
+  stable or changes materially as signal-to-noise degrades. This design also
+  encourages careful evaluation with repeated random seeds and test-set metrics
+  that capture both average error and fit quality.
+
+  A credible implementation should compare at least one representative for each
+  family (e.g., RandomForestRegressor for bagging, GradientBoostingRegressor
+  for boosting, StackingRegressor with diverse base estimators for stacking),
+  include a single-model baseline, and report RMSE-centered conclusions. Since
+  this benchmark is intended for fast autonomous execution, datasets and models
+  must remain small enough to run in minutes on one CPU core while still
+  producing clear numeric contrasts across noise regimes.
+
+  The key outcome is not reproducing a known paper result but determining when
+  each ensemble strategy is most robust for noisy non-linear regression in a
+  constrained practical setting.
+
+  *How does predictive robustness (RMSE) of bagging, boosting, and stacking change as regression noise increases on small non-linear benchmarks?*
+
+hypotheses:
+  - id: H1
+    statement: "On the high-noise synthetic dataset (noise >= 25), bagging (RandomForestRegressor) achieves lower mean RMSE than boosting (GradientBoostingRegressor), averaged over at least 5 seeds."
+    measurable: true
+  - id: H2
+    statement: "On at least 2 of 3 evaluated datasets, at least one ensemble method (bagging, boosting, or stacking) improves RMSE by >= 10% relative to the single DecisionTreeRegressor baseline."
+    measurable: true
+  - id: H3
+    statement: "StackingRegressor attains the best (lowest) mean RMSE on at least 1 of 3 datasets while not being worst on more than 1 dataset, averaged over at least 5 seeds."
+    measurable: true
+
+experiment_design:
+  research_question: "How does RMSE performance of bagging, boosting, and stacking vary with increasing noise in non-linear regression tasks under CPU-only constraints?"
+  conditions:
+    - name: "decision_tree_baseline"
+      description: "Single DecisionTreeRegressor (max_depth tuned over a small grid) as non-ensemble baseline."
+    - name: "bagging_random_forest"
+      description: "RandomForestRegressor representing bagging-style averaging across trees."
+    - name: "boosting_gbrt"
+      description: "GradientBoostingRegressor representing sequential boosting on residuals."
+    - name: "stacking_heterogeneous"
+      description: "StackingRegressor with base learners {kNN regressor, ridge regression, shallow random forest} and a linear meta-regressor."
+  baselines:
+    - "decision_tree_baseline is the single-model baseline"
+    - "bagging_random_forest serves as a strong classical ensemble reference"
+  metrics:
+    - name: "rmse"
+      direction: "minimize"
+      description: "Root Mean Squared Error on held-out test split, averaged over seeds."
+    - name: "mae"
+      direction: "minimize"
+      description: "Mean Absolute Error on held-out test split, averaged over seeds."
+    - name: "r2"
+      direction: "maximize"
+      description: "Coefficient of determination on held-out test split, averaged over seeds."
+  datasets:
+    - name: "friedman1_low_noise"
+      source: "sklearn.datasets.make_friedman1 (n_samples=1200, n_features=10, noise=1.0)"
+    - name: "friedman1_high_noise"
+      source: "sklearn.datasets.make_friedman1 (n_samples=1200, n_features=10, noise=30.0)"
+    - name: "diabetes"
+      source: "sklearn.datasets.load_diabetes"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML02.json"

tasks/ml/manifests/ML03.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T03 — Gradient-free optimization on non-convex benchmark functions
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML03
+title: "CPU comparison of Nelder-Mead, Powell, and CMA-ES on non-convex functions"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Gradient-free optimization methods are widely used when derivatives are
+  unavailable, unreliable, or expensive to compute. In low-to-medium
+  dimensional non-convex landscapes, direct-search methods such as
+  Nelder-Mead and Powell are frequently used because they are easy to call
+  through scipy.optimize. Population-based approaches such as CMA-ES can be
+  more robust to local minima and ill-conditioning, but they introduce extra
+  hyperparameters and potentially higher per-iteration cost.
+
+  A CPU-bounded benchmark can clarify practical trade-offs by evaluating these
+  methods under a fixed function-evaluation budget on standard synthetic
+  objective functions with known global minima. Rather than asking which method
+  is universally best, the relevant question is whether one method reaches
+  better objective values more reliably within the same budget and how much
+  runtime overhead that reliability costs.
+
+  A credible experiment should evaluate at least three optimization conditions
+  (Nelder-Mead, Powell, CMA-ES) across multiple non-convex test functions,
+  using repeated random starts and common stopping/budget rules. Reporting only
+  final objective value is incomplete; runtime and success-rate-to-threshold
+  should be included to capture both quality and efficiency. Statistical
+  summaries over seeds are required because single runs are high variance.
+
+  The resulting evidence should produce explicit pass/fail verdicts for each
+  hypothesis: whether CMA-ES improves best-found objective value, whether
+  Powell offers faster wall-clock convergence than CMA-ES at similar budgets,
+  and whether Nelder-Mead underperforms on multimodal landscapes. This keeps
+  the study measurable and feasible within a short single-core runtime window.
+
+  *Under equal function-evaluation budgets on CPU, which gradient-free optimizer (Nelder-Mead, Powell, CMA-ES) gives the best trade-off between final objective quality, runtime, and success rate on non-convex benchmark functions?*
+
+hypotheses:
+  - id: H1
+    statement: "CMA-ES achieves lower mean best objective value than both Nelder-Mead and Powell on at least 2 of 3 benchmark functions, averaged over >=10 random starts with the same evaluation budget."
+    measurable: true
+  - id: H2
+    statement: "Powell has lower median wall-clock runtime than CMA-ES on at least 2 of 3 benchmark functions while finishing within the same function-evaluation cap."
+    measurable: true
+  - id: H3
+    statement: "On the multimodal Rastrigin function, Nelder-Mead attains a lower success rate (fraction of runs reaching f(x) <= 1e-2) than CMA-ES by at least 0.20 absolute."
+    measurable: true
+
+experiment_design:
+  research_question: "Under equal evaluation budgets, how do Nelder-Mead, Powell, and CMA-ES compare on objective quality, runtime, and success probability for non-convex synthetic objectives?"
+  conditions:
+    - name: "nelder_mead"
+      description: "scipy.optimize.minimize with method='Nelder-Mead', random initial point per run, maxfev budget enforced."
+    - name: "powell"
+      description: "scipy.optimize.minimize with method='Powell', same initialization protocol and maxfev cap."
+    - name: "cma_es"
+      description: "CMA-ES implementation (lightweight numpy/scipy variant) with population updates under the same total function-evaluation budget."
+    - name: "random_search_baseline"
+      description: "Uniform random search within bounded domain using the same number of objective evaluations as other methods."
+  baselines:
+    - "random_search_baseline as a budget-matched non-adaptive baseline"
+  metrics:
+    - name: "primary_metric"
+      direction: "minimize"
+      description: "Mean best objective value found at termination (lower is better), aggregated over random starts."
+    - name: "median_runtime_sec"
+      direction: "minimize"
+      description: "Median wall-clock runtime per run in seconds for each (method, function)."
+    - name: "success_rate_eps"
+      direction: "maximize"
+      description: "Fraction of runs reaching objective value <= 1e-2 by termination."
+  datasets:
+    - name: "rastrigin_10d"
+      source: "synthetic numpy implementation of Rastrigin function in 10 dimensions"
+    - name: "rosenbrock_10d"
+      source: "synthetic numpy implementation of Rosenbrock function in 10 dimensions"
+    - name: "ackley_10d"
+      source: "synthetic numpy implementation of Ackley function in 10 dimensions"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 480
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML03.json"

tasks/ml/manifests/ML04.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T04 — Feature scaling choices for k-nearest neighbors under distribution shift
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML04
+title: "Effect of standard, min-max, and robust scaling on KNN classification"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  K-nearest neighbors (KNN) is highly sensitive to feature scale because
+  distance calculations implicitly weight dimensions by their numeric ranges.
+  In practice, preprocessing decisions such as StandardScaler, MinMaxScaler,
+  or RobustScaler can alter neighborhood structure enough to change
+  classification performance as much as model hyperparameters do. Despite this,
+  scaling choice is often treated as a default rather than as an experimental
+  factor.
+
+  The core methodological question is not whether scaling helps versus no
+  scaling (it usually does), but whether specific scaling families are better
+  matched to particular data distributions. Standard scaling assumes roughly
+  Gaussian-like behavior, min-max scaling compresses to bounded ranges and can
+  be strongly affected by extrema, and robust scaling centers/scales by median
+  and IQR to reduce outlier influence.
+
+  A CPU-friendly, credible benchmark should compare these scalers with a fixed
+  KNN classifier across multiple datasets that differ in outlier prevalence and
+  marginal feature distributions, while averaging over multiple train/test
+  splits. Because KNN is lightweight on small-to-medium sklearn datasets, the
+  study can include both predictive quality and stability metrics without
+  exceeding strict runtime constraints.
+
+  The experiment should report test accuracy as the primary outcome, plus
+  macro-F1 and split-to-split variability, and include at least one outlier-
+  heavy synthetic dataset to stress robust scaling behavior. The goal is to
+  produce falsifiable conclusions about when scaler choice materially changes
+  KNN outcomes.
+
+  *How does the choice among standard, min-max, and robust feature scaling change KNN classification performance and stability across datasets with different distribution and outlier characteristics?*
+
+hypotheses:
+  - id: H1
+    statement: "RobustScaler + KNN achieves higher mean test accuracy than StandardScaler + KNN on the outlier-heavy synthetic dataset by at least 0.03 absolute accuracy, averaged over ≥5 random splits."
+    measurable: true
+  - id: H2
+    statement: "Across the evaluated datasets, at least one real dataset shows an absolute test-accuracy gap of ≥0.02 between the best and worst of {StandardScaler, MinMaxScaler, RobustScaler}, indicating scaler choice materially affects KNN performance."
+    measurable: true
+  - id: H3
+    statement: "No_scaling baseline is not the top-accuracy condition on at least 2 of 3 datasets when compared against the three scaling methods with the same KNN hyperparameters."
+    measurable: true
+
+experiment_design:
+  research_question: "How does scaler choice (standard, min-max, robust) affect KNN classification accuracy and stability across datasets with different feature distributions and outlier profiles?"
+  conditions:
+    - name: "no_scaling_knn"
+      description: "KNeighborsClassifier with fixed k and distance metric on raw features; control condition."
+    - name: "standard_scaler_knn"
+      description: "Pipeline(StandardScaler, KNeighborsClassifier) with the same KNN hyperparameters as control."
+    - name: "minmax_scaler_knn"
+      description: "Pipeline(MinMaxScaler, KNeighborsClassifier) with identical KNN settings."
+    - name: "robust_scaler_knn"
+      description: "Pipeline(RobustScaler, KNeighborsClassifier) with identical KNN settings."
+  baselines:
+    - "no_scaling_knn is the preprocessing-free baseline"
+  metrics:
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Mean held-out accuracy over at least 5 random stratified splits per dataset."
+    - name: "macro_f1"
+      direction: "maximize"
+      description: "Macro-averaged F1 on held-out data, averaged over splits."
+    - name: "accuracy_std"
+      direction: "minimize"
+      description: "Standard deviation of test accuracy across random splits as a stability indicator."
+  datasets:
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "synthetic_outlier_classification"
+      source: "sklearn.datasets.make_classification with injected feature outliers on a subset of samples"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 180
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML04.json"

tasks/ml/manifests/ML05.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T05 — Dimensionality reduction for preserving cluster structure
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML05
+title: "Dimensionality reduction methods for preserving cluster structure in synthetic high-dimensional data"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Dimensionality reduction is frequently used before visualization and
+  clustering, but different methods optimize different objectives and can
+  distort neighborhood and global geometry in distinct ways. PCA is linear and
+  emphasizes variance preservation, while t-SNE and UMAP are nonlinear manifold
+  methods designed to preserve local structure. In practice, users often infer
+  cluster quality from 2D embeddings without checking whether separability in
+  the embedding reflects true high-dimensional structure.
+
+  A CPU-scale investigation can probe this mismatch by generating synthetic
+  datasets where ground-truth cluster labels are known and controllable. By
+  varying cluster overlap, anisotropy, and manifold geometry (e.g., Gaussian
+  blobs, anisotropic transforms, and noisy moons embedded in high dimensions),
+  one can evaluate whether each reducer preserves cluster structure under
+  different regimes instead of relying on a single toy example.
+
+  A credible experiment should compare PCA, t-SNE, and UMAP under a common
+  pipeline: reduce to 2 dimensions, then score structure using metrics such as
+  silhouette score and adjusted Rand index after a fixed clustering backend
+  (e.g., KMeans with true k). Runtime should also be tracked because nonlinear
+  methods may yield better structure at substantially higher computational cost,
+  which matters in constrained environments.
+
+  The goal is not to claim a universally best reducer, but to quantify tradeoffs
+  between structure preservation and efficiency in settings where intrinsic
+  geometry differs. This produces actionable guidance for method choice based on
+  data regime rather than defaults.
+
+  *Which dimensionality reduction method (PCA, t-SNE, or UMAP) best preserves cluster structure across distinct synthetic high-dimensional regimes when evaluated by silhouette score, ARI, and runtime?*
+
+hypotheses:
+  - id: H1
+    statement: "On nonlinear manifold data (noisy two-moons embedded into 50D), at least one nonlinear reducer (t-SNE or UMAP) achieves silhouette_score at least 0.10 higher than PCA after 2D reduction and KMeans clustering, averaged over >=5 seeds."
+    measurable: true
+  - id: H2
+    statement: "On linearly separable Gaussian blobs in 50D, PCA achieves silhouette_score within 0.05 of the best nonlinear method (t-SNE or UMAP), averaged over >=5 seeds."
+    measurable: true
+  - id: H3
+    statement: "PCA has the lowest mean wall-clock reduction time and is at least 3x faster than both t-SNE and UMAP on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Which reducer best balances cluster-structure preservation and compute cost across linear and nonlinear synthetic high-dimensional datasets?"
+  conditions:
+    - name: "pca_2d"
+      description: "PCA with n_components=2 using sklearn.decomposition.PCA."
+    - name: "tsne_2d"
+      description: "t-SNE with n_components=2, perplexity in [20, 40], learning_rate='auto', init='pca'."
+    - name: "umap_2d"
+      description: "UMAP-like reducer: use umap.UMAP(n_components=2, n_neighbors in [15, 30], min_dist in [0.0, 0.3]) if available; if unavailable, use sklearn.manifold.SpectralEmbedding(n_components=2) as a documented fallback proxy."
+    - name: "identity_no_reduction"
+      description: "Baseline that applies KMeans directly in the original high-dimensional space (no dimensionality reduction)."
+  baselines:
+    - "identity_no_reduction as no-DR clustering baseline"
+    - "pca_2d as linear DR baseline"
+  metrics:
+    - name: "silhouette_score"
+      direction: "maximize"
+      description: "Silhouette score computed on the 2D embedding (or original space for identity baseline) using predicted KMeans labels; primary metric."
+    - name: "ari"
+      direction: "maximize"
+      description: "Adjusted Rand Index between KMeans labels and known synthetic ground-truth labels."
+    - name: "fit_transform_time_sec"
+      direction: "minimize"
+      description: "Wall-clock time for dimensionality reduction fit_transform only (seconds)."
+  datasets:
+    - name: "gaussian_blobs_50d"
+      source: "sklearn.datasets.make_blobs with 4 centers, n_features=50, cluster_std=1.2"
+    - name: "anisotropic_blobs_50d"
+      source: "make_blobs in 10D expanded/projection to 50D then linear anisotropic transform"
+    - name: "embedded_moons_50d"
+      source: "sklearn.datasets.make_moons(noise=0.08) embedded into 50D by random linear projection plus Gaussian noise"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 600
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML05.json"

tasks/ml/manifests/ML06.yaml

@@ -0,0 +1,100 @@
+# ============================================================================
+# T06 — Adaptive learning-rate schedules for logistic-regression convergence
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML06
+title: "Adaptive learning-rate schedules for logistic-regression convergence on binary classification"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Logistic regression is often treated as a solved baseline, yet its practical
+  training behavior still depends strongly on optimization details. In
+  first-order methods, the learning-rate schedule can dominate time-to-quality:
+  a constant step size may converge slowly or oscillate, while adaptive
+  schedules can accelerate early progress or stabilize late optimization.
+  Because logistic regression has a smooth convex objective, it offers a clean
+  setting to compare schedules without confounds from deep-network
+  non-convexity.
+
+  This topic studies four schedule families in a unified mini-batch gradient
+  descent implementation: step decay, exponential decay, cosine annealing, and
+  cosine warm restarts. The focus is not raw final accuracy alone, but
+  optimization efficiency measured by epochs needed to reach a predefined
+  validation-loss threshold, plus robustness across datasets and random seeds.
+  A fixed-learning-rate baseline is required so any claimed speedup is
+  attributable to schedule design.
+
+  A credible CPU-scale experiment should run on multiple binary classification
+  datasets from sklearn, with feature standardization, identical model
+  parameterization, and synchronized stopping criteria. Reporting should include
+  convergence epochs, final validation log loss, and test ROC-AUC so that faster
+  convergence is not achieved at the expense of generalization. Since warm
+  restarts can re-increase the learning rate, trajectory logging is important to
+  verify that restart behavior is actually implemented.
+
+  The key decision criterion is whether adaptive schedules reduce optimization
+  effort in a consistent, measurable way. The study should conclude with clear
+  per-hypothesis verdicts and practical limitations (dataset size, threshold
+  choice, and sensitivity to initial learning rate).
+
+  *Do adaptive learning-rate schedules (especially cosine variants) reduce logistic-regression convergence epochs versus fixed/monotone decay schedules while maintaining comparable binary-classification performance?*
+
+hypotheses:
+  - id: H1
+    statement: "Cosine annealing or cosine warm restarts achieves at least 20% lower mean convergence_epochs than fixed learning rate on at least 2 of 3 evaluated binary datasets, averaged over >=5 seeds."
+    measurable: true
+  - id: H2
+    statement: "Among adaptive schedules (step decay, exponential decay, cosine annealing, warm restarts), the best schedule's final validation log loss is no worse than 0.01 absolute compared with the worst schedule on each dataset (i.e., speed differences exceed quality differences)."
+    measurable: true
+  - id: H3
+    statement: "Warm restarts reaches convergence in fewer epochs than monotone exponential decay on at least 2 of 3 datasets, without reducing test ROC-AUC by more than 0.01 absolute on those datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Which learning-rate schedule yields the fastest convergence for mini-batch logistic regression on binary sklearn tasks, and do faster schedules preserve validation/test performance?"
+  conditions:
+    - name: "fixed_lr"
+      description: "Mini-batch logistic regression trained with constant learning rate (baseline)."
+    - name: "step_decay"
+      description: "Learning rate multiplied by gamma at fixed epoch milestones (e.g., every 20 epochs)."
+    - name: "exponential_decay"
+      description: "Learning rate updated each epoch as lr_t = lr0 * exp(-k * t)."
+    - name: "cosine_annealing"
+      description: "Learning rate follows cosine decay from lr0 to lr_min over total epochs without restarts."
+    - name: "cosine_warm_restarts"
+      description: "Cosine schedule with periodic restarts (SGDR-style) resetting to higher learning rate at restart boundaries."
+  baselines:
+    - "fixed_lr is the primary optimization baseline"
+    - "exponential_decay is a monotone adaptive baseline for comparison to warm restarts"
+  metrics:
+    - name: "convergence_epochs"
+      direction: "minimize"
+      description: "Epoch index at which validation log loss first drops below a preset threshold (or max_epochs if never reached), averaged over seeds."
+    - name: "val_log_loss"
+      direction: "minimize"
+      description: "Final validation logistic loss after training, mean over seeds."
+    - name: "test_roc_auc"
+      direction: "maximize"
+      description: "ROC-AUC on held-out test set, mean over seeds."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "synthetic_binary_medium"
+      source: "sklearn.datasets.make_classification (n_samples=2000, n_features=30, n_informative=12, class_sep=1.0)"
+    - name: "synthetic_binary_noisy"
+      source: "sklearn.datasets.make_classification (n_samples=2500, n_features=40, n_informative=10, flip_y=0.08, class_sep=0.7)"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML06.json"

tasks/ml/manifests/ML07.yaml

@@ -0,0 +1,92 @@
+# ============================================================================
+# T07 — Text feature extraction strategies for linear document classifiers
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML07
+title: "Text feature extraction trade-offs: TF-IDF vs count vs hashing with NB/SVM"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Sparse bag-of-words pipelines remain strong baselines for document
+  classification, especially under strict CPU and latency constraints.
+  Three common feature extractors — CountVectorizer, TfidfVectorizer, and
+  HashingVectorizer — define different bias/variance and efficiency trade-offs.
+  Count vectors preserve raw frequency signals useful for generative models;
+  TF-IDF often improves linear margin classifiers by downweighting common
+  tokens; hashing avoids vocabulary construction and can reduce memory and
+  fit-time at the cost of collisions.
+
+  Classifier choice interacts strongly with the representation. Multinomial
+  Naive Bayes (MNB) is typically paired with nonnegative count-like features,
+  while linear SVM variants often benefit from TF-IDF normalization. In
+  practical workflows, teams frequently need to choose between slightly better
+  macro-F1 and much faster runtime. This makes a pure "best score" benchmark
+  incomplete unless paired with timing and robustness checks.
+
+  A credible CPU-scale study should compare multiple extractor+classifier
+  combinations on at least two document datasets available directly through
+  sklearn (or trivially synthesized text), use fixed train/test splits with
+  repeated seeds where relevant, and report both predictive quality and
+  efficiency metrics. The experiment should explicitly test whether TF-IDF +
+  linear SVM provides the strongest macro-F1 while hashing offers meaningful
+  speed advantages with limited degradation.
+
+  *Does TF-IDF with linear SVM deliver the best macro-F1 on sklearn text benchmarks, and is HashingVectorizer a worthwhile speed/accuracy trade-off versus vocabulary-based features?*
+
+hypotheses:
+  - id: H1
+    statement: "TF-IDF + linear SVM achieves the highest macro-F1 among implemented conditions on at least 2 of 3 evaluated text datasets."
+    measurable: true
+  - id: H2
+    statement: "HashingVectorizer-based pipelines reduce vectorization+fit wall-clock time by at least 20% versus the corresponding TF-IDF pipeline on at least 2 of 3 datasets."
+    measurable: true
+  - id: H3
+    statement: "Multinomial Naive Bayes with count vectors attains macro-F1 within 0.05 absolute of TF-IDF + linear SVM on at least 1 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Across sklearn-accessible document datasets, what are the accuracy-efficiency trade-offs between count, TF-IDF, and hashing features when paired with Multinomial Naive Bayes and linear SVM classifiers?"
+  conditions:
+    - name: "count_mnb"
+      description: "CountVectorizer (word unigrams, min_df=2) + MultinomialNB(alpha=1.0)."
+    - name: "tfidf_mnb"
+      description: "TfidfVectorizer (word unigrams, min_df=2, norm=l2) + MultinomialNB(alpha=1.0)."
+    - name: "tfidf_linear_svm"
+      description: "TfidfVectorizer (word unigrams, min_df=2, norm=l2) + LinearSVC(C=1.0)."
+    - name: "hashing_linear_svm"
+      description: "HashingVectorizer (word unigrams, n_features=2^18, alternate_sign=False) + LinearSVC(C=1.0)."
+  baselines:
+    - "count_mnb as the classic sparse-text baseline"
+    - "tfidf_mnb as a same-classifier feature-ablation baseline"
+  metrics:
+    - name: "macro_f1"
+      direction: "maximize"
+      description: "Macro-averaged F1 score on held-out test split."
+    - name: "accuracy"
+      direction: "maximize"
+      description: "Overall test accuracy for comparability with prior text benchmarks."
+    - name: "fit_predict_time_sec"
+      direction: "minimize"
+      description: "End-to-end wall-clock time for vectorization, model fit, and test prediction."
+  datasets:
+    - name: "20newsgroups_4class"
+      source: "sklearn.datasets.fetch_20newsgroups with 4 categories (subset=train/test, remove=(\"headers\",\"footers\",\"quotes\"))"
+    - name: "20newsgroups_8class"
+      source: "sklearn.datasets.fetch_20newsgroups with 8 categories (subset=train/test, remove=(\"headers\",\"footers\",\"quotes\"))"
+    - name: "synthetic_topic_docs"
+      source: "Trivially synthesized corpus via templated sentence generation for 4 classes, 2000 documents total"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 600
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML07.json"

tasks/ml/manifests/ML08.yaml

@@ -0,0 +1,102 @@
+# ============================================================================
+# T08 — Class-imbalance handling strategies for binary classification
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML08
+title: "Impact of imbalance-handling strategies on sklearn binary classifiers"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Class imbalance is one of the most common practical failure modes in binary
+  classification: a model can achieve high raw accuracy by predicting the
+  majority class while missing the minority class almost entirely. In
+  scikit-learn workflows, practitioners often choose among simple data-level
+  resampling (random oversampling, random undersampling), algorithm-level
+  weighting (class_weight="balanced"), or synthetic methods like SMOTE.
+  However, these strategies trade off minority recall, precision, and overall
+  discrimination differently, and their behavior varies with dataset geometry.
+
+  A compact CPU-friendly study can still be rigorous if it compares multiple
+  imbalance-handling strategies under the same base learner and preprocessing
+  pipeline, evaluates with metrics appropriate for skewed labels (balanced
+  accuracy, F1, PR-AUC), and averages over several random seeds. The focus is
+  not on maximizing one benchmark score but on understanding whether methods
+  that improve minority detection do so consistently across datasets, and at
+  what precision cost.
+
+  To keep implementation feasible in a short autonomous run, the experiment can
+  use sklearn-native datasets and synthetic imbalance generated from a standard
+  source dataset, with a shared train/test protocol and fixed model family
+  (e.g., logistic regression). SMOTE can be implemented in a lightweight way
+  using sklearn.neighbors to interpolate minority-neighbor pairs, avoiding any
+  external dependency.
+
+  The key analytical goal is to test whether weighted learning or synthetic
+  oversampling provides the most reliable gain in balanced accuracy over a
+  no-treatment baseline, and whether naive random oversampling tends to hurt
+  precision-oriented metrics relative to class weighting.
+
+  *Which imbalance-handling strategy (SMOTE, random oversampling, class weights, undersampling) gives the most consistent balanced-accuracy improvement for binary sklearn classification under a fixed model and protocol?*
+
+hypotheses:
+  - id: H1
+    statement: "At least one of {class_weight_balanced, smote} achieves higher balanced_accuracy than no_handling by >=0.03 absolute on at least 2 of 3 datasets (mean over >=5 seeds)."
+    measurable: true
+  - id: H2
+    statement: "Random undersampling yields the highest minority recall among compared strategies on at least 2 of 3 datasets, but does not achieve the best PR-AUC on those same datasets."
+    measurable: true
+  - id: H3
+    statement: "Random oversampling does not outperform class_weight_balanced in balanced_accuracy on more than 1 of 3 datasets (seed-averaged comparison)."
+    measurable: true
+
+experiment_design:
+  research_question: "Which imbalance-handling strategy provides the most consistent gains in balanced accuracy for binary classification with fixed sklearn models under controlled train/test splits?"
+  conditions:
+    - name: "no_handling"
+      description: "Baseline logistic regression trained on imbalanced data with no resampling and default class weights."
+    - name: "class_weight_balanced"
+      description: "Same logistic regression with class_weight='balanced'."
+    - name: "random_oversampling"
+      description: "Training-set minority class randomly oversampled with replacement to match majority count."
+    - name: "random_undersampling"
+      description: "Training-set majority class randomly undersampled without replacement to match minority count."
+    - name: "smote_k5"
+      description: "Training-set minority class augmented with synthetic samples via kNN interpolation (k=5) until class balance is reached."
+  baselines:
+    - "no_handling is the primary baseline"
+    - "class_weight_balanced serves as an algorithm-level baseline against data-level resampling"
+  metrics:
+    - name: "balanced_accuracy"
+      direction: "maximize"
+      description: "Mean of sensitivity and specificity on held-out test split, averaged over seeds."
+    - name: "minority_recall"
+      direction: "maximize"
+      description: "Recall for the positive/minority class on test data."
+    - name: "average_precision"
+      direction: "maximize"
+      description: "Area under precision-recall curve (PR-AUC / AP) on test data."
+    - name: "f1"
+      direction: "maximize"
+      description: "Binary F1 score for the minority class at default threshold 0.5."
+  datasets:
+    - name: "breast_cancer_imbalanced"
+      source: "sklearn.datasets.load_breast_cancer with induced imbalance by downsampling positive class in training folds"
+    - name: "make_classification_90_10"
+      source: "sklearn.datasets.make_classification(n_samples=4000, n_features=20, weights=[0.9,0.1], random_state=seed)"
+    - name: "make_classification_95_05"
+      source: "sklearn.datasets.make_classification(n_samples=5000, n_features=25, weights=[0.95,0.05], class_sep=1.0, random_state=seed)"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 360
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML08.json"

tasks/ml/manifests/ML09.yaml

@@ -0,0 +1,101 @@
+# ============================================================================
+# T09 — Comparing Bayesian optimization vs grid search vs random search for
+# hyperparameter tuning of random forests on UCI benchmark datasets
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML09
+title: "Bayesian optimization vs grid vs random search for random-forest tuning"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Hyperparameter tuning often dominates practical AutoML performance, yet in
+  CPU-limited settings the search strategy itself can matter as much as the
+  model family. For random forests, common tunables (number of trees,
+  max_depth, min_samples_split, min_samples_leaf, max_features) define a mixed
+  discrete-continuous space where exhaustive grids can be wasteful,
+  random search can be surprisingly strong, and Bayesian optimization can use
+  surrogate modeling to focus evaluations on promising regions.
+
+  A concise benchmark can test this tradeoff by fixing a single model class
+  (RandomForestClassifier), fixed CV protocol, and comparable evaluation budget
+  across search methods. Grid search should use a compact but principled grid;
+  random search should sample from the same parameter ranges; Bayesian
+  optimization can be implemented with sklearn's Gaussian-process based
+  optimizer (e.g., BayesSearchCV if available is not allowed here, so a custom
+  lightweight GP + acquisition loop or scipy/sklearn surrogate approach is
+  expected). The key is fairness: equal or near-equal numbers of objective
+  evaluations and identical data splits.
+
+  Since this benchmark must run in under 15 minutes on one CPU core,
+  datasets should be small-to-medium sklearn/UCI-style classification tasks
+  already available in sklearn (e.g., wine, breast_cancer, digits). The
+  analysis should report best cross-validation score, held-out test accuracy,
+  and search efficiency (score gain per evaluation or wall-clock).
+  Multi-seed repeats are desirable to reduce noise, but can be kept modest to
+  remain within runtime limits.
+
+  The experiment should answer whether Bayesian optimization delivers better
+  best-found configurations under tight evaluation budgets, or whether simpler
+  baselines (especially random search) are effectively equivalent for these
+  tabular tasks. Interpretation should explicitly separate optimization quality
+  from compute cost.
+
+  *Under a fixed small evaluation budget, does Bayesian optimization find better random-forest hyperparameters than grid and random search on sklearn UCI-style classification benchmarks?*
+
+hypotheses:
+  - id: H1
+    statement: "With an equal budget of 20 hyperparameter evaluations per dataset, Bayesian optimization achieves higher mean best_cv_score than random search on at least 2 of 3 datasets."
+    measurable: true
+  - id: H2
+    statement: "Random search matches or exceeds grid search in best_cv_score on at least 2 of 3 datasets under the same maximum number of evaluated configurations."
+    measurable: true
+  - id: H3
+    statement: "The mean held-out test accuracy of the best Bayesian-tuned model is at least 0.5 percentage points higher than the best grid-tuned model on at least 1 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Under a fixed evaluation budget, which tuning strategy (Bayesian optimization, grid search, random search) yields the best random-forest cross-validation and test performance on sklearn UCI-style datasets?"
+  conditions:
+    - name: "grid_search_budget20"
+      description: "Grid search over a compact predefined grid for RandomForestClassifier, capped at 20 evaluated configurations per dataset."
+    - name: "random_search_budget20"
+      description: "Random search over matched hyperparameter ranges with 20 sampled configurations per dataset."
+    - name: "bayes_opt_budget20"
+      description: "Bayesian optimization using a Gaussian-process surrogate and acquisition-guided proposals for 20 total evaluations per dataset."
+    - name: "default_rf"
+      description: "Untuned RandomForestClassifier with sklearn default hyperparameters as a reference baseline."
+  baselines:
+    - "default_rf as no-tuning baseline"
+    - "grid_search_budget20 as classical tuning baseline"
+  metrics:
+    - name: "best_cv_score"
+      direction: "maximize"
+      description: "Best mean 3-fold cross-validation accuracy discovered by each search method."
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Accuracy on a held-out test split using the best hyperparameters found by each method."
+    - name: "search_time_sec"
+      direction: "minimize"
+      description: "Wall-clock time spent in hyperparameter search per method and dataset."
+  datasets:
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 780
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML09.json"

tasks/ml/manifests/ML10.yaml

@@ -0,0 +1,98 @@
+# ============================================================================
+# T10 — Cross-validation strategy effects on small-sample model selection
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML10
+title: "Cross-validation strategy reliability for small-sample model selection"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  With small datasets, model selection can become highly sensitive to the
+  validation protocol. Practitioners often choose among k-fold,
+  stratified k-fold, repeated k-fold, and leave-one-out cross-validation
+  (LOOCV), but these strategies trade off bias, variance, and compute in
+  different ways. A method that appears best under one CV scheme may not be
+  best under another, especially when class balance is imperfect and sample
+  counts are low.
+
+  This topic studies CV strategy as the independent variable while keeping
+  candidate models fixed and lightweight. The key quantity is estimation bias:
+  the gap between CV-estimated score used for model selection and the realized
+  test score of the selected model on a held-out test set. In small-sample
+  settings, minimizing this gap is often more important than maximizing raw CV
+  score, because over-optimistic selection can produce brittle deployments.
+
+  A credible CPU-scale experiment should evaluate multiple sklearn datasets
+  reduced to small training sizes, run several random seeds, and compare at
+  least four CV strategies under the same model grid and scoring rule. The
+  study should report not only estimation bias but also variance across seeds
+  and model-selection stability. Baselines should include standard k-fold and
+  stratified k-fold; repeated stratified k-fold and LOOCV serve as contrasting
+  alternatives with different variance/compute characteristics.
+
+  The goal is not to prove one universally superior protocol, but to quantify
+  when more elaborate CV schemes improve reliability enough to justify their
+  cost under strict CPU budgets.
+
+  *Which cross-validation strategy yields the lowest and most stable model-selection estimation bias for small-sample classification tasks under a fixed candidate-model grid?*
+
+hypotheses:
+  - id: H1
+    statement: "RepeatedStratifiedKFold (5 folds × 3 repeats) achieves lower mean absolute estimation bias than plain KFold (5 folds) on at least 2 of 3 evaluated small-sample datasets, averaged over ≥5 random seeds."
+    measurable: true
+  - id: H2
+    statement: "StratifiedKFold (5 folds) yields lower across-seed standard deviation of selected-model test accuracy than plain KFold (5 folds) on at least 2 of 3 datasets."
+    measurable: true
+  - id: H3
+    statement: "LOOCV does not outperform RepeatedStratifiedKFold in mean absolute estimation bias by more than 0.01 absolute score on any dataset, while requiring greater wall-clock time."
+    measurable: true
+
+experiment_design:
+  research_question: "For small-sample classification model selection, how do KFold, StratifiedKFold, RepeatedStratifiedKFold, and LOOCV compare in estimation bias, stability, and compute cost?"
+  conditions:
+    - name: "kfold_5"
+      description: "Model selection via 5-fold KFold (shuffle=True) on the training split."
+    - name: "stratified_kfold_5"
+      description: "Model selection via 5-fold StratifiedKFold (shuffle=True) on the training split."
+    - name: "repeated_stratified_5x3"
+      description: "Model selection via RepeatedStratifiedKFold with 5 folds and 3 repeats."
+    - name: "loocv"
+      description: "Model selection via Leave-One-Out cross-validation on the training split."
+  baselines:
+    - "kfold_5 as the non-stratified baseline"
+    - "stratified_kfold_5 as the standard small-sample baseline"
+  metrics:
+    - name: "estimation_bias"
+      direction: "minimize"
+      description: "Absolute difference between best CV mean score and held-out test score of the selected model, aggregated over seeds."
+    - name: "selected_model_test_accuracy"
+      direction: "maximize"
+      description: "Held-out test accuracy of the model selected by each CV strategy, mean over seeds."
+    - name: "selection_stability"
+      direction: "maximize"
+      description: "Fraction of seeds selecting the modal hyperparameter configuration (higher = more stable)."
+    - name: "wall_clock_sec"
+      direction: "minimize"
+      description: "Runtime per CV strategy over all seeds and datasets on CPU."
+  datasets:
+    - name: "breast_cancer_small"
+      source: "sklearn.datasets.load_breast_cancer (subsample training set to n<=120)"
+    - name: "wine_small"
+      source: "sklearn.datasets.load_wine (subsample training set to n<=100)"
+    - name: "synthetic_imbalanced_small"
+      source: "sklearn.datasets.make_classification (n_samples=160, weights=[0.75,0.25], class_sep tuned for moderate difficulty)"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 600
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML10.json"

tasks/ml/manifests/ML11.yaml

@@ -0,0 +1,98 @@
+# ============================================================================
+# T11 — Unsupervised outlier detection on tabular datasets with injected anomalies
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML11
+title: "Benchmarking classical unsupervised outlier detectors under controlled anomaly injection"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Classical unsupervised anomaly-detection methods remain widely used in
+  production tabular pipelines because they are interpretable enough,
+  CPU-efficient, and available in standard libraries. Yet their relative
+  behavior can change sharply as anomaly prevalence and feature scaling vary.
+  IsolationForest is often robust in heterogeneous feature spaces, while
+  distance/density methods like LocalOutlierFactor can be strong when local
+  neighborhoods are informative but may degrade in higher dimensions.
+
+  A practical benchmark should avoid external downloads and still provide
+  realistic control over anomaly rate. One reliable strategy is to start from
+  clean sklearn classification datasets and inject synthetic anomalies into the
+  test split by sampling uniformly beyond feature-wise quantile ranges of the
+  training data. This yields known binary ground truth for evaluation while
+  preserving a realistic inlier distribution.
+
+  A credible CPU-scale study here should compare IsolationForest,
+  LocalOutlierFactor (novelty mode), OneClassSVM, and EllipticEnvelope under a
+  shared preprocessing pipeline (e.g., StandardScaler), evaluate ROC-AUC and
+  PR-AUC across at least two datasets and multiple seeds, and include a simple
+  baseline such as random scoring. Since contamination is rarely known exactly
+  in practice, the benchmark should also probe at least two injected anomaly
+  rates.
+
+  The main decision-relevant question is not whether any method can detect easy
+  anomalies, but which method is consistently best across datasets and
+  contamination levels under strict CPU constraints. A concise, quantitative
+  hypothesis-driven comparison is therefore the goal.
+
+  *Which classical unsupervised outlier detector is most robust across tabular datasets and injected anomaly rates when measured by ROC-AUC and PR-AUC under a single-core CPU budget?*
+
+hypotheses:
+  - id: H1
+    statement: "IsolationForest achieves higher mean ROC-AUC than OneClassSVM on at least 2 of 3 datasets when averaged over anomaly rates {0.05, 0.10} and ≥3 seeds."
+    measurable: true
+  - id: H2
+    statement: "At anomaly rate 0.10, LocalOutlierFactor attains the best PR-AUC (rank 1) on at least 1 of 3 datasets."
+    measurable: true
+  - id: H3
+    statement: "Each implemented detector (IsolationForest, LocalOutlierFactor, OneClassSVM, EllipticEnvelope) outperforms a random-score baseline in ROC-AUC by at least 0.10 on the pooled mean across all evaluated datasets/rates."
+    measurable: true
+
+experiment_design:
+  research_question: "Which among IsolationForest, LocalOutlierFactor, OneClassSVM, and EllipticEnvelope is most robust across datasets and injected anomaly rates for unsupervised outlier detection under CPU constraints?"
+  conditions:
+    - name: "isolation_forest"
+      description: "sklearn IsolationForest with contamination set to injected rate, 200 trees, default feature subsampling."
+    - name: "local_outlier_factor_novelty"
+      description: "sklearn LocalOutlierFactor in novelty=True mode, n_neighbors in [20, 35], contamination set to injected rate."
+    - name: "one_class_svm_rbf"
+      description: "sklearn OneClassSVM with RBF kernel, nu matched to injected rate, gamma='scale'."
+    - name: "elliptic_envelope"
+      description: "sklearn EllipticEnvelope with contamination set to injected rate and robust covariance estimation."
+    - name: "random_score_baseline"
+      description: "Baseline assigning i.i.d. uniform random anomaly scores to test points."
+  baselines:
+    - "random_score_baseline"
+    - "one_class_svm_rbf as a classical margin-based reference"
+  metrics:
+    - name: "roc_auc"
+      direction: "maximize"
+      description: "ROC-AUC of anomaly scores against injected ground-truth labels on the test set, averaged over seeds."
+    - name: "pr_auc"
+      direction: "maximize"
+      description: "Average precision / PR-AUC for anomaly class on the test set, averaged over seeds."
+    - name: "fit_predict_time_sec"
+      direction: "minimize"
+      description: "Wall-clock seconds for model fit plus test scoring per (dataset, rate, seed), reported as mean."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 480
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML11.json"

tasks/ml/manifests/ML12.yaml

@@ -0,0 +1,96 @@
+# ============================================================================
+# T12 — Clustering algorithm comparison on synthetic geometric datasets
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML12
+title: "Comparing clustering algorithms on synthetic non-convex and anisotropic shapes"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Clustering algorithms encode very different geometric assumptions. K-means
+  prefers spherical, equal-variance groups; agglomerative clustering can adapt
+  to hierarchical structure depending on linkage; DBSCAN finds dense regions
+  and can recover non-convex shapes while labeling outliers as noise; spectral
+  clustering can separate manifolds when graph affinity is appropriate.
+  Because these assumptions interact strongly with data geometry, a single
+  benchmark score often hides systematic failure modes.
+
+  Synthetic datasets are ideal for a fast but meaningful comparison because we
+  can generate known structures with ground-truth labels: intertwined moons,
+  concentric circles, and anisotropic Gaussian blobs. These patterns expose
+  strengths and weaknesses that are difficult to isolate in real-world data.
+  With labeled synthetic data, external clustering metrics like adjusted rand
+  index (ARI) and normalized mutual information (NMI) directly quantify
+  partition recovery quality.
+
+  A credible CPU-scale study should compare several algorithms under a shared
+  protocol: standardized inputs, fixed train-size generation per seed, and
+  explicit handling of methods that output noise labels. It should include at
+  least one centroid baseline (k-means) and one density method (DBSCAN), then
+  test whether non-linear methods systematically outperform centroid methods on
+  non-convex shapes while avoiding large regressions on anisotropic blobs.
+
+  The practical goal is not to crown one universally best algorithm, but to
+  establish data-shape-dependent guidance. If shape complexity changes which
+  method wins, practitioners should choose clustering tools by geometric prior
+  rather than habit.
+
+  *How strongly does dataset geometry (non-convex vs anisotropic) determine which clustering algorithm achieves the best ground-truth agreement under a fixed CPU-budget protocol?*
+
+hypotheses:
+  - id: H1
+    statement: "On the two non-convex datasets (moons and circles), at least one of {DBSCAN, SpectralClustering} achieves mean ARI that is at least 0.15 higher than k-means, averaged over ≥5 seeds."
+    measurable: true
+  - id: H2
+    statement: "On the anisotropic-blobs dataset, agglomerative clustering (ward linkage) achieves ARI greater than or equal to k-means in the seed-averaged result (difference >= 0.00)."
+    measurable: true
+  - id: H3
+    statement: "No single algorithm ranks first in ARI on all evaluated datasets; i.e., the best-ARI method differs on at least one dataset from the global winner."
+    measurable: true
+
+experiment_design:
+  research_question: "How strongly does dataset geometry (non-convex vs anisotropic) determine which clustering algorithm achieves the best agreement with ground-truth labels under a fixed CPU-budget protocol?"
+  conditions:
+    - name: "kmeans_k2_or_k3"
+      description: "KMeans baseline with n_clusters set to dataset ground-truth class count; n_init>=10."
+    - name: "agglomerative_ward"
+      description: "AgglomerativeClustering with ward linkage and n_clusters equal to ground-truth class count."
+    - name: "dbscan_tuned_eps"
+      description: "DBSCAN with eps selected from a small grid per dataset (e.g., 0.1 to 0.5) and fixed min_samples (e.g., 5)."
+    - name: "spectral_rbf"
+      description: "SpectralClustering with RBF affinity and n_clusters equal to ground-truth class count."
+  baselines:
+    - "kmeans_k2_or_k3 as centroid-based baseline"
+    - "agglomerative_ward as hierarchical baseline"
+  metrics:
+    - name: "adjusted_rand_score"
+      direction: "maximize"
+      description: "Adjusted Rand Index against ground-truth labels, averaged over seeds."
+    - name: "normalized_mutual_info"
+      direction: "maximize"
+      description: "Normalized Mutual Information against ground-truth labels."
+    - name: "silhouette_score"
+      direction: "maximize"
+      description: "Internal cohesion/separation score computed from predicted labels (excluding noise-only failures)."
+  datasets:
+    - name: "moons"
+      source: "sklearn.datasets.make_moons (n_samples~800, noise~0.08)"
+    - name: "circles"
+      source: "sklearn.datasets.make_circles (n_samples~800, noise~0.06, factor~0.5)"
+    - name: "anisotropic_blobs"
+      source: "sklearn.datasets.make_blobs followed by linear transformation to create anisotropy"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML12.json"

tasks/ml/manifests/ML13.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T13 — Kernel choice effects for Gaussian Process regression
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML13
+title: "Evaluating kernel choices for Gaussian Process regression on synthetic 1-D and 5-D functions"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Gaussian Process (GP) regression performance depends strongly on the kernel,
+  which encodes assumptions about function smoothness and local structure.
+  Practitioners often default to the RBF kernel because it is easy to optimize
+  and widely available, but this choice can underperform when the target
+  function has roughness patterns that are better matched by Matern families or
+  when polynomial trends dominate part of the signal. Since kernel
+  misspecification can be subtle, a short benchmark should compare predictive
+  uncertainty quality as well as point error.
+
+  A CPU-scale study can be done entirely with sklearn GaussianProcessRegressor
+  on synthetic datasets where the data-generating process is controlled. In 1-D,
+  a mildly nonstationary smooth function with additive Gaussian noise can expose
+  differences in extrapolation and uncertainty tails. In 5-D, a mixed
+  sinusoidal-plus-quadratic target can stress anisotropy and interactions while
+  remaining fast enough for repeated fitting under multiple random seeds.
+
+  To make claims measurable, the experiment should evaluate at least four kernel
+  choices (RBF, polynomial, Matern-3/2, Matern-5/2) under the same train/test
+  splits, optimizer restarts, and noise model assumptions. Because the topic's
+  key metric is negative log-likelihood, uncertainty calibration quality should
+  be primary; RMSE and R^2 provide supporting evidence for point prediction.
+  Seed averaging is important because synthetic splits and optimizer
+  initialization can change marginal likelihood optimization outcomes.
+
+  The core question is whether smoother kernels (RBF, Matern-5/2) consistently
+  outperform rougher or mismatched kernels (Matern-3/2, polynomial) on test NLL
+  across both low- and moderate-dimensional synthetic tasks, and whether ranking
+  by NLL aligns with ranking by RMSE.
+
+  *Which GP kernel among RBF, polynomial, Matern-3/2, and Matern-5/2 yields the best uncertainty-aware generalization (lowest test NLL) on synthetic 1-D and 5-D noisy regression functions?*
+
+hypotheses:
+  - id: H1
+    statement: "Across the two synthetic datasets, at least one of {RBF, Matern-5/2} achieves the lowest mean test NLL on each dataset (averaged over >=5 seeds)."
+    measurable: true
+  - id: H2
+    statement: "The polynomial kernel has mean test NLL at least 10% worse than the best kernel on at least 1 of the 2 datasets."
+    measurable: true
+  - id: H3
+    statement: "Kernel ranking by mean test NLL and by mean RMSE is not identical on at least 1 dataset, indicating uncertainty quality and point error disagree."
+    measurable: true
+
+experiment_design:
+  research_question: "How do RBF, polynomial, Matern-3/2, and Matern-5/2 kernels compare for Gaussian Process regression on synthetic 1-D and 5-D noisy functions when judged primarily by test NLL?"
+  conditions:
+    - name: "gp_rbf"
+      description: "GaussianProcessRegressor with ConstantKernel * RBF + WhiteKernel, hyperparameters optimized by log-marginal-likelihood."
+    - name: "gp_poly_deg2"
+      description: "GaussianProcessRegressor with ConstantKernel * DotProduct^2 (implemented via polynomial feature mapping + DotProduct kernel) + WhiteKernel as a polynomial-trend proxy."
+    - name: "gp_matern32"
+      description: "GaussianProcessRegressor with ConstantKernel * Matern(nu=1.5) + WhiteKernel."
+    - name: "gp_matern52"
+      description: "GaussianProcessRegressor with ConstantKernel * Matern(nu=2.5) + WhiteKernel."
+  baselines:
+    - "gp_rbf as the common default-kernel baseline"
+    - "gp_poly_deg2 as a mismatched-trend baseline"
+  metrics:
+    - name: "test_nll"
+      direction: "minimize"
+      description: "Mean negative log predictive density on held-out test data using GP predictive mean and variance, averaged over seeds."
+    - name: "rmse"
+      direction: "minimize"
+      description: "Root mean squared error on test targets, averaged over seeds."
+    - name: "r2"
+      direction: "maximize"
+      description: "Coefficient of determination on test targets, averaged over seeds."
+  datasets:
+    - name: "synthetic_1d_sinmix"
+      source: "Generated with numpy: x in [-3,3], y = sin(2x) + 0.3x + epsilon, epsilon~N(0, 0.15^2)."
+    - name: "synthetic_5d_mixed"
+      source: "Generated with numpy: X in [-2,2]^5, y = sin(x1) + 0.5*x2^2 - 0.7*x3 + 0.3*x4*x5 + epsilon, epsilon~N(0, 0.2^2)."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 360
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML13.json"

tasks/ml/manifests/ML14.yaml

@@ -0,0 +1,96 @@
+# ============================================================================
+# T14 — Comparing conformal prediction procedures on heteroscedastic regression
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML14
+title: "Comparing split-conformal, Mondrian conformal, and CQR-style intervals on heteroscedastic regression"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Conformal prediction provides finite-sample coverage guarantees with minimal
+  assumptions, making it attractive for uncertainty quantification in practical
+  regression. However, standard split-conformal intervals are often globally
+  calibrated and can be inefficient when noise is heteroscedastic: they may
+  over-cover low-noise regions and under-cover high-noise regions. This
+  motivates conditional variants such as Mondrian conformal (group-conditional
+  calibration) and CQR-style approaches that adapt interval width through
+  quantile modeling before conformal correction.
+
+  In small, CPU-only settings, one can still run a meaningful comparison by
+  using sklearn-compatible regressors and synthetic data where heteroscedastic
+  structure is controlled. A rigorous setup should evaluate not only marginal
+  coverage but also efficiency (mean interval width) and conditional behavior
+  (coverage gap across strata of predicted difficulty or input regions). These
+  diagnostics reveal when methods achieve nominal coverage by producing overly
+  wide intervals versus genuinely adapting to varying noise.
+
+  A credible benchmark should include at least one homoscedastic control and at
+  least one heteroscedastic dataset, then compare split-conformal, a Mondrian
+  variant (binning by an auxiliary score such as |x| or predicted scale), and a
+  CQR-style method based on lower/upper quantile regressors plus conformal
+  adjustment. Repeated random splits (multiple seeds) are needed because
+  conformal procedures can vary with calibration sample composition.
+
+  The key question is whether adaptive procedures can reduce conditional
+  miscoverage while maintaining near-target marginal coverage and reasonable
+  interval width under strict runtime constraints.
+
+  *Do Mondrian and CQR-style conformal procedures achieve lower conditional coverage error than vanilla split-conformal at comparable marginal coverage on heteroscedastic regression tasks?*
+
+hypotheses:
+  - id: H1
+    statement: "On heteroscedastic datasets, at least one adaptive method (Mondrian or CQR-style) attains a lower absolute coverage gap to the 90% target than split-conformal by at least 0.02 on at least 2 of 3 datasets, averaged over \u22655 seeds."
+    measurable: true
+  - id: H2
+    statement: "At matched target 90% coverage, CQR-style conformal yields mean interval width no larger than split-conformal on at least 2 of 3 datasets (difference \u2264 0.00)."
+    measurable: true
+  - id: H3
+    statement: "Mondrian conformal reduces worst-bin conditional miscoverage (max over 4 bins of |bin_coverage-0.90|) by at least 0.03 versus split-conformal on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Do adaptive conformal procedures (Mondrian, CQR-style) improve coverage quality and efficiency over split-conformal for heteroscedastic regression under a fixed 90% target coverage?"
+  conditions:
+    - name: "split_conformal_rf"
+      description: "Vanilla split-conformal regression using RandomForestRegressor point predictor and absolute residual conformity scores on a held-out calibration split."
+    - name: "mondrian_conformal_rf_4bins"
+      description: "Mondrian split-conformal with the same base predictor, calibration and test points partitioned into 4 bins by an auxiliary score (e.g., fitted value or |x0| proxy), with per-bin quantiles."
+    - name: "cqr_gbr"
+      description: "CQR-style method using GradientBoostingRegressor quantile models (alpha=0.05, 0.95) and split-conformal correction on calibration residuals of quantile bands."
+    - name: "naive_quantile_no_conformal"
+      description: "Non-conformal baseline using raw quantile regression interval [q0.05, q0.95] without conformal adjustment."
+  baselines:
+    - "split_conformal_rf is the primary conformal baseline"
+    - "naive_quantile_no_conformal is the non-conformal baseline"
+  metrics:
+    - name: "coverage_gap"
+      direction: "minimize"
+      description: "Absolute difference between empirical marginal coverage and target 0.90 on the test split, averaged over seeds."
+    - name: "mean_interval_width"
+      direction: "minimize"
+      description: "Average prediction interval width on test samples, averaged over seeds."
+    - name: "worst_bin_miscoverage"
+      direction: "minimize"
+      description: "Maximum across 4 bins of absolute deviation between bin coverage and 0.90, measuring conditional coverage disparity."
+  datasets:
+    - name: "hetero_sine"
+      source: "synthetic: y = sin(2\u03c0x) + (0.1 + 0.5|x|)\u03b5, x~Uniform(-1,1), n\u22482000"
+    - name: "hetero_friedman1"
+      source: "synthetic: sklearn.datasets.make_friedman1 with multiplicative noise scale depending on x0"
+    - name: "diabetes"
+      source: "sklearn.datasets.load_diabetes (tabular regression benchmark)"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML14.json"

tasks/ml/manifests/ML15.yaml

@@ -0,0 +1,96 @@
+# ============================================================================
+# T15 — Filter vs embedded feature selection under irrelevant-feature injection
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML15
+title: "Filter-based vs embedded L1 feature selection with injected noise features"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Feature selection on tabular classification problems is often presented as a
+  choice between simple univariate filters and embedded sparse models.
+  Univariate filters such as chi-squared, ANOVA F-score, and mutual
+  information are computationally cheap and easy to apply in a pipeline, but
+  they evaluate each feature independently and can miss interactions. Embedded
+  L1-regularized logistic regression performs selection during model fitting,
+  potentially yielding feature subsets that better align with predictive
+  structure when many irrelevant variables are present.
+
+  A practical stress test is to inject synthetic irrelevant features into
+  otherwise standard benchmark datasets. This setup creates controlled feature
+  dilution while preserving original labels and base difficulty. Under such
+  dilution, methods that can ignore noise should maintain accuracy and avoid
+  selecting many synthetic features. Conversely, methods sensitive to spurious
+  univariate associations may degrade as noise dimension increases.
+
+  A credible CPU-only study should compare at least three filter selectors
+  (mutual_info_classif, chi2, f_classif) and an embedded L1 selector in a
+  shared downstream classifier pipeline, evaluate on multiple sklearn
+  classification datasets, and report both predictive performance and selection
+  quality. Including multiple random seeds is important because both noise
+  injection and train/test splits can change apparent selector rankings.
+
+  The key aim is not to reproduce a single published table, but to determine
+  whether embedded sparsity offers robustness advantages over filters as
+  irrelevant dimensions grow, and whether that robustness generalizes across
+  datasets with different feature scales and class structures.
+
+  *Does embedded L1-logistic feature selection retain predictive performance and reject injected irrelevant features more reliably than univariate filter methods under controlled feature-noise injection?*
+
+hypotheses:
+  - id: H1
+    statement: "With 5x injected irrelevant features (relative to original feature count), L1-logistic embedded selection achieves higher mean test_accuracy than each filter method (mutual_info_classif, chi2, f_classif) on at least 2 of 3 datasets, averaged over >=5 seeds."
+    measurable: true
+  - id: H2
+    statement: "At fixed selected-feature budget k=20 (or all features if p<20), L1-logistic selects a lower fraction of injected irrelevant features than the average of the three filter methods on all evaluated datasets."
+    measurable: true
+  - id: H3
+    statement: "From no-injection to 5x-injection settings, mean test_accuracy drop for L1-logistic is <= 3 percentage points on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Does embedded L1-logistic feature selection outperform univariate filter methods in robustness to injected irrelevant features on sklearn tabular classification datasets?"
+  conditions:
+    - name: "filter_mutual_info_k20"
+      description: "SelectKBest(mutual_info_classif, k=20 or p if p<20) followed by LogisticRegression classifier."
+    - name: "filter_chi2_k20"
+      description: "MinMax scaling to nonnegative domain, then SelectKBest(chi2, k=20 or p if p<20) followed by LogisticRegression classifier."
+    - name: "filter_f_classif_k20"
+      description: "SelectKBest(f_classif, k=20 or p if p<20) followed by LogisticRegression classifier."
+    - name: "embedded_l1_logistic"
+      description: "L1-penalized LogisticRegression (saga/liblinear) used as embedded selector; keep top-20 by absolute coefficient magnitude (or nonzero if <=20), retrain LogisticRegression on selected subset."
+  baselines:
+    - "filter_f_classif_k20 as a standard univariate linear-statistic baseline"
+    - "filter_mutual_info_k20 as a nonlinear dependency baseline"
+  metrics:
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Held-out test accuracy averaged over >=5 seeds for each dataset and injection level."
+    - name: "noise_feature_selection_rate"
+      direction: "minimize"
+      description: "Fraction of selected features that come from injected irrelevant columns."
+    - name: "accuracy_drop_0x_to_5x_pp"
+      direction: "minimize"
+      description: "Absolute percentage-point drop in test_accuracy from 0x to 5x injection."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML15.json"

tasks/ml/manifests/ML16.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T16 — Comparing stochastic and adversarial bandit algorithms under drift
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML16
+title: "Bandit algorithm robustness under stationary and drifting reward regimes"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Multi-armed bandit algorithms encode different assumptions about reward
+  generation and non-stationarity. Epsilon-greedy is simple and adaptable with
+  persistent exploration; UCB1 is optimism-driven and often strong in
+  stationary stochastic settings; Thompson sampling can be highly sample
+  efficient when model assumptions are matched; Exp3 is designed for
+  adversarial settings and may trade off stochastic efficiency for robustness.
+  In small CPU-constrained studies, these differences are often discussed
+  qualitatively but not tested with matched synthetic environments.
+
+  A meaningful benchmark should evaluate both Bernoulli and Gaussian reward
+  arms, because posterior/model assumptions and noise scale can materially
+  change outcomes. It should also include stationary and drifting regimes:
+  methods tuned for fixed means can degrade when the best arm changes over
+  time. Regret, not just final reward, is the right primary metric because it
+  captures online learning efficiency throughout the horizon.
+
+  A credible experiment here implements the four algorithms with consistent
+  action/reward interfaces, runs repeated simulations over multiple seeds, and
+  reports cumulative regret trajectories and endpoint statistics. To keep the
+  study CPU-friendly, horizons and arm counts should be modest (e.g., 5-10 arms,
+  500-2000 steps), with vectorized numpy updates where possible.
+
+  The key analytical value is comparative: identify where UCB1/Thompson excel
+  in stationary stochastic settings, and whether epsilon-greedy or Exp3 is more
+  resilient under drift. The objective is not reproducing a single paper, but
+  testing algorithm-environment fit under controlled synthetic shifts.
+
+  *How do epsilon-greedy, UCB1, Thompson sampling, and Exp3 trade off cumulative regret across Bernoulli vs Gaussian arms and stationary vs drifting reward regimes?*
+
+hypotheses:
+  - id: H1
+    statement: "In stationary Bernoulli environments, either UCB1 or Thompson sampling achieves at least 10% lower mean cumulative regret than epsilon-greedy at horizon T on at least 2 of 3 datasets (averaged over >=20 seeds)."
+    measurable: true
+  - id: H2
+    statement: "In drifting environments, epsilon-greedy or Exp3 achieves lower mean cumulative regret than UCB1 on at least 2 of 3 datasets at horizon T (averaged over >=20 seeds)."
+    measurable: true
+  - id: H3
+    statement: "Across all evaluated datasets, no single algorithm is best (lowest mean cumulative regret) on every dataset, indicating environment-dependent performance."
+    measurable: true
+
+experiment_design:
+  research_question: "How do epsilon-greedy, UCB1, Thompson sampling, and Exp3 compare in cumulative regret across stationary versus drifting synthetic bandit tasks with Bernoulli and Gaussian rewards?"
+  conditions:
+    - name: "epsilon_greedy_eps0.1"
+      description: "Epsilon-greedy with constant epsilon=0.1 and sample-mean value estimates."
+    - name: "ucb1"
+      description: "UCB1 with exploration bonus sqrt(2 log t / n_a)."
+    - name: "thompson_sampling"
+      description: "Thompson sampling using Beta-Bernoulli updates for Bernoulli arms and Gaussian posterior sampling (known variance assumption) for Gaussian arms."
+    - name: "exp3_gamma0.07"
+      description: "Exp3 with gamma=0.07 and importance-weighted reward estimates."
+  baselines:
+    - "epsilon_greedy_eps0.1 as simple stochastic baseline"
+    - "ucb1 as canonical optimism baseline"
+  metrics:
+    - name: "cumulative_regret"
+      direction: "minimize"
+      description: "Mean cumulative regret at final horizon T relative to oracle best arm per round, averaged over seeds."
+    - name: "instantaneous_regret_auc"
+      direction: "minimize"
+      description: "Area under per-round regret curve over time (lower is better)."
+    - name: "best_arm_selection_rate"
+      direction: "maximize"
+      description: "Fraction of rounds selecting the current optimal arm (for drift: time-varying optimum)."
+  datasets:
+    - name: "bernoulli_stationary_k10"
+      source: "synthetic: 10 Bernoulli arms with fixed means sampled in [0.05, 0.95] and sorted gap >=0.05"
+    - name: "bernoulli_drift_k10"
+      source: "synthetic: 10 Bernoulli arms with piecewise-constant means; best arm switches at predefined changepoints"
+    - name: "gaussian_drift_k8"
+      source: "synthetic: 8 Gaussian arms (sigma=1) with linearly drifting means and periodic rank reversals"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML16.json"

tasks/ml/manifests/ML17.yaml

@@ -0,0 +1,97 @@
+# ============================================================================
+# T17 — Comparing topic models (LDA, NMF, LSA) on 20newsgroups subsets
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML17
+title: "Topic-model comparison on small 20newsgroups subsets: LDA vs NMF vs LSA"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Topic modeling methods are often compared using either intrinsic quality
+  measures (such as topic coherence) or extrinsic utility (such as how well
+  document representations align with known labels), but the two views can
+  disagree. Latent Dirichlet Allocation (LDA), Non-negative Matrix
+  Factorization (NMF), and Latent Semantic Analysis (LSA/SVD) each induce
+  different assumptions over term-document structure and may therefore rank
+  differently depending on metric choice.
+
+  On CPU-constrained benchmarks, a practical study should use a compact text
+  corpus and a controlled preprocessing pipeline so that runtime stays short
+  while still yielding meaningful distinctions. A small subset of
+  20newsgroups is ideal because it has accessible labels for external
+  clustering evaluation and enough lexical diversity for coherence analysis.
+  Using multiple subset difficulties (well-separated vs more confusable
+  categories) helps test robustness of conclusions.
+
+  A credible experiment compares LDA, NMF, and LSA under matched topic counts
+  and vectorization settings, then reports both c_v-like coherence and
+  adjusted rand index (ARI) from document-cluster assignments against true
+  newsgroup labels. Because exact c_v implementations are not native in
+  sklearn, an explicit approximation based on sliding-window co-occurrence
+  and normalized PMI should be documented and applied consistently across
+  methods. Repeated runs for stochastic models are needed to avoid
+  over-interpreting single-seed noise.
+
+  The core question is whether better intrinsic coherence implies better
+  label alignment on small real-world corpora, and which model offers the
+  best trade-off under strict CPU budgets.
+
+  *Do LDA, NMF, and LSA produce different rankings on coherence versus ARI on small 20newsgroups subsets, and does NMF provide the strongest coherence-structure trade-off under CPU limits?*
+
+hypotheses:
+  - id: H1
+    statement: "NMF achieves higher mean coherence_cv than LSA on at least 2 of 3 evaluated 20newsgroups subsets when using the same number of topics and preprocessing pipeline."
+    measurable: true
+  - id: H2
+    statement: "Across the three methods (LDA, NMF, LSA), the method with the highest coherence_cv is also the highest-ARI method on no more than 1 of the 3 subsets, indicating weak metric agreement."
+    measurable: true
+  - id: H3
+    statement: "LDA achieves mean ARI at least 0.03 higher than LSA on at least 2 of 3 subsets when document clusters are obtained by argmax topic assignment."
+    measurable: true
+
+experiment_design:
+  research_question: "How do LDA, NMF, and LSA compare on coherence_cv versus document-cluster ARI over small 20newsgroups subsets, and do intrinsic/extrinsic rankings diverge?"
+  conditions:
+    - name: "lda_k4"
+      description: "LatentDirichletAllocation with n_components=4, max_iter=20, learning_method='batch', CountVectorizer features, repeated over 3 seeds."
+    - name: "nmf_k4"
+      description: "NMF with n_components=4, init='nndsvda', solver='cd', max_iter=300 on TF-IDF features, repeated over 3 seeds."
+    - name: "lsa_k4"
+      description: "TruncatedSVD (LSA) with n_components=4 on TF-IDF features; document-topic embeddings clustered via argmax after non-negative shift or via KMeans(k=4) with fixed seed."
+    - name: "nmf_k6_sensitivity"
+      description: "NMF sensitivity condition with n_components=6 to test topic-count robustness for coherence and ARI trends."
+  baselines:
+    - "lsa_k4 as a linear-algebra baseline without probabilistic topic assumptions"
+    - "lda_k4 as a probabilistic-topic baseline"
+  metrics:
+    - name: "coherence_cv"
+      direction: "maximize"
+      description: "Approximate c_v coherence computed from top words per topic using sliding-window co-occurrence and NPMI aggregation, averaged across topics and seeds."
+    - name: "ari"
+      direction: "maximize"
+      description: "Adjusted Rand Index between predicted document clusters (topic argmax or equivalent) and true newsgroup labels."
+    - name: "runtime_sec"
+      direction: "minimize"
+      description: "Wall-clock training + inference time per condition/dataset cell on single CPU core."
+  datasets:
+    - name: "20ng_easy4"
+      source: "sklearn.datasets.fetch_20newsgroups with categories=['sci.space','rec.autos','comp.graphics','talk.politics.misc']"
+    - name: "20ng_related4"
+      source: "sklearn.datasets.fetch_20newsgroups with categories=['comp.graphics','comp.os.ms-windows.misc','comp.sys.ibm.pc.hardware','comp.sys.mac.hardware']"
+    - name: "20ng_science4"
+      source: "sklearn.datasets.fetch_20newsgroups with categories=['sci.space','sci.med','sci.electronics','sci.crypt']"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 720
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML17.json"

tasks/ml/manifests/ML18.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T18 — Post-hoc probability calibration for sklearn classifiers
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML18
+title: "Post-hoc calibration methods for sklearn classifiers on tabular benchmarks"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Many widely used scikit-learn classifiers optimize discrimination rather
+  than probability quality. Random forests, gradient boosting models, and
+  RBF-kernel SVMs can achieve strong accuracy while output probabilities are
+  miscalibrated, especially on small or moderately imbalanced tabular data.
+  This matters in decision settings where thresholds, risk ranking, or cost-
+  sensitive actions depend on reliable confidence estimates.
+
+  Post-hoc calibration methods are attractive because they can be layered onto
+  existing models without retraining the base learner. Platt scaling fits a
+  sigmoid map, isotonic regression fits a non-parametric monotone map, and
+  temperature scaling applies a single-parameter logit rescaling (implemented
+  for binary tasks via optimization on held-out logits/probabilities). These
+  methods differ in flexibility and overfitting risk, so their relative value
+  may depend on classifier family and dataset size.
+
+  A credible CPU-scale study should compare uncalibrated outputs versus at
+  least three calibrators across multiple sklearn tabular datasets, using a
+  proper train/calibration/test protocol and repeated seeds. Because this is a
+  calibration-centric question, expected calibration error (ECE) and log loss
+  should be primary metrics, with accuracy used as a guardrail to ensure that
+  calibration does not degrade classification utility.
+
+  The goal is not to reproduce a single paper result but to test whether any
+  one calibrator is consistently superior across heterogeneous models and
+  datasets under tight compute constraints. The key uncertainty is whether
+  method ranking is stable enough to justify a default choice.
+
+  *Which post-hoc calibration method (Platt, isotonic, temperature) most reliably improves ECE across RF/GBM/SVM-RBF on small tabular sklearn benchmarks without materially harming accuracy?*
+
+hypotheses:
+  - id: H1
+    statement: "At least one post-hoc calibrator (Platt, isotonic, or temperature) reduces ECE by at least 10% relative to the uncalibrated model for at least 2 of 3 classifiers on at least 2 of 3 datasets, averaged over >=3 seeds."
+    measurable: true
+  - id: H2
+    statement: "Isotonic regression achieves lower mean ECE than Platt scaling on at least 2 of 3 datasets when calibration-set size is >=150 samples."
+    measurable: true
+  - id: H3
+    statement: "Applying post-hoc calibration changes test accuracy by no more than 1.0 absolute percentage point (mean over seeds) for each classifier-dataset pair."
+    measurable: true
+
+experiment_design:
+  research_question: "Which post-hoc calibration method (Platt, isotonic, temperature) most consistently improves probability calibration for RF/GBM/SVM-RBF on sklearn tabular datasets while preserving accuracy?"
+  conditions:
+    - name: "uncalibrated"
+      description: "Base classifier probabilities/decision scores used directly with no calibration."
+    - name: "platt_scaling"
+      description: "Sigmoid calibration fit on a held-out calibration split (CalibratedClassifierCV with method='sigmoid' or equivalent)."
+    - name: "isotonic_regression"
+      description: "Isotonic calibration fit on a held-out calibration split (CalibratedClassifierCV with method='isotonic' or equivalent)."
+    - name: "temperature_scaling"
+      description: "Single temperature parameter optimized on calibration split to minimize NLL; applied to logits/scores before probability mapping."
+  baselines:
+    - "uncalibrated outputs are the primary baseline"
+    - "platt_scaling serves as a classic parametric calibration baseline"
+  metrics:
+    - name: "ece"
+      direction: "minimize"
+      description: "Expected Calibration Error (10-15 bins) on the held-out test split, averaged over seeds."
+    - name: "log_loss"
+      direction: "minimize"
+      description: "Negative log-likelihood on test probabilities."
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Classification accuracy on held-out test split to verify discrimination is preserved."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits_binary"
+      source: "sklearn.datasets.load_digits with target transformed to binary (e.g., digit<5 vs >=5)"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 480
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML18.json"

tasks/ml/manifests/ML19.yaml

@@ -0,0 +1,95 @@
+# ============================================================================
+# T19 — Semi-supervised strategies on small tabular datasets
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML19
+title: "Comparing graph- and pseudo-label-based semi-supervised learning on small tabular datasets"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Semi-supervised learning is attractive when labels are scarce but unlabeled
+  examples are cheap. In sklearn, LabelPropagation and LabelSpreading provide
+  graph-based transductive approaches, while SelfTrainingClassifier offers a
+  pseudo-labeling wrapper around a standard supervised base learner. These
+  methods rely on different assumptions: graph smoothness over neighborhoods
+  versus confidence-thresholded iterative self-labeling. On small tabular
+  datasets, their relative behavior can change quickly as the labeled fraction
+  drops from moderate (20%) to very low (10%).
+
+  A useful CPU-scale study should compare these methods under the same splits,
+  feature preprocessing, and random seeds, with explicit control of labeled
+  fraction. Because the key claim of semi-supervision is label efficiency, the
+  experiment should include a supervised-only baseline trained on the same
+  labeled subset and evaluated on a fixed held-out test set. This makes gains
+  attributable to unlabeled-data usage rather than favorable splits.
+
+  The study should report not only test accuracy but also balanced accuracy and
+  macro-F1, since small tabular datasets can have class imbalance and accuracy
+  alone may hide minority-class degradation. Repeating runs across several seeds
+  is necessary because which points are labeled strongly affects outcomes at low
+  label rates.
+
+  Practically, the benchmark should stay lightweight: sklearn datasets only,
+  no downloads, and model choices that complete within minutes on a single CPU
+  core. The focus is not SOTA performance but robust relative comparisons across
+  label regimes and datasets.
+
+  *How do LabelPropagation, LabelSpreading, and SelfTrainingClassifier compare in label-efficiency versus a supervised-only baseline when only 10–20% of training labels are available on small sklearn tabular datasets?*
+
+hypotheses:
+  - id: H1
+    statement: "At 10% labeled data, at least one semi-supervised method (LabelPropagation, LabelSpreading, or SelfTrainingClassifier) achieves test accuracy at least 3 percentage points higher than the supervised-only baseline on at least 2 of 3 datasets, averaged over ≥5 seeds."
+    measurable: true
+  - id: H2
+    statement: "Across datasets at 10% labeled data, LabelSpreading with RBF kernel attains mean test accuracy greater than or equal to LabelPropagation with RBF kernel in at least 2 of 3 datasets (seed-averaged)."
+    measurable: true
+  - id: H3
+    statement: "For SelfTrainingClassifier, increasing labeled fraction from 10% to 20% improves mean test accuracy by at least 1 absolute percentage point on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "How do graph-based (LabelPropagation/LabelSpreading) and pseudo-labeling (SelfTrainingClassifier) methods compare to supervised-only training under 10% and 20% labeled-data regimes on small tabular benchmarks?"
+  conditions:
+    - name: "supervised_logreg_labeled_only"
+      description: "LogisticRegression trained only on the labeled subset of the training split; unlabeled points ignored."
+    - name: "label_propagation_rbf"
+      description: "LabelPropagation with RBF kernel; unlabeled training labels set to -1 and transductive predictions used for test inference."
+    - name: "label_spreading_rbf"
+      description: "LabelSpreading with RBF kernel under the same masked-label protocol."
+    - name: "self_training_logreg"
+      description: "SelfTrainingClassifier wrapping LogisticRegression with confidence threshold (e.g., 0.8), fit on mixed labeled/unlabeled training data."
+  baselines:
+    - "supervised_logreg_labeled_only is the primary baseline"
+    - "10% labeled regime serves as a lower-label baseline versus 20% within each method"
+  metrics:
+    - name: "test_accuracy"
+      direction: "maximize"
+      description: "Classification accuracy on held-out test split, averaged over seeds."
+    - name: "balanced_accuracy"
+      direction: "maximize"
+      description: "Balanced accuracy on held-out test split, averaged over seeds."
+    - name: "macro_f1"
+      direction: "maximize"
+      description: "Macro-averaged F1 score on held-out test split, averaged over seeds."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML19.json"

tasks/ml/manifests/ML20.yaml

@@ -0,0 +1,102 @@
+# ============================================================================
+# T20 — Benchmarking classical time-series forecasters on synthetic seasonal series
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML20
+title: "Classical forecaster robustness on synthetic seasonal time series"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Classical univariate forecasting methods remain widely used because they are
+  interpretable and lightweight, yet their comparative behavior can flip under
+  different data-generating regimes. AR models often perform well when dynamics
+  are mostly autoregressive and weakly seasonal; ARIMA can absorb trend and
+  differencing structure; SARIMAX can explicitly represent seasonal lag effects;
+  ETS captures error-trend-seasonal decomposition; and Theta is often a strong
+  low-variance baseline. In small CPU-constrained settings, practitioners need
+  practical guidance about which method is robust to changing signal-to-noise
+  and seasonality strength.
+
+  A synthetic benchmark is appropriate here because we can precisely control
+  trend slope, seasonal amplitude, and observation noise while keeping compute
+  modest. By generating multiple families of monthly-like series with known
+  periodicity (e.g., period 12), we can evaluate whether model rankings are
+  stable or regime-dependent. This avoids overfitting conclusions to one
+  real-world dataset and enables direct stress testing under high-noise versus
+  high-seasonality conditions.
+
+  A credible study should compare at least four of {AR, ARIMA, SARIMAX, ETS,
+  Theta} on 2–3 synthetic dataset families, using rolling-origin or holdout
+  forecasts with a fixed horizon. It should report scale-robust error metrics
+  (especially sMAPE), include at least one baseline (e.g., seasonal naive), and
+  aggregate over multiple random seeds. The analysis should explicitly test
+  whether one method is consistently best overall or whether the best choice
+  depends on data regime.
+
+  The key outcome is actionable selection logic: if seasonal structure is strong
+  and noise is low, do seasonal/state-space methods dominate; and if noise is
+  high, do simpler methods become competitive? These are concrete decisions an
+  autonomous forecasting pipeline can apply when only limited data and CPU are
+  available.
+
+  *How do AR, ARIMA, SARIMAX, ETS, and Theta compare in sMAPE robustness across synthetic seasonal series with controlled trend, seasonality amplitude, and noise levels?*
+
+hypotheses:
+  - id: H1
+    statement: "SARIMAX or ETS achieves the lowest mean sMAPE on at least 2 of 3 synthetic dataset families when seasonality amplitude is high (amplitude >= 8) and noise is low (sigma <= 1.0), averaged over >=5 seeds."
+    measurable: true
+  - id: H2
+    statement: "Under high-noise settings (sigma >= 3.0), the sMAPE gap between the best advanced method (ARIMA/SARIMAX/ETS/Theta) and a seasonal_naive baseline is <= 5 percentage points on at least 2 of 3 dataset families."
+    measurable: true
+  - id: H3
+    statement: "No single method among {AR, ARIMA, SARIMAX, ETS, Theta} ranks first in mean sMAPE on all evaluated dataset families, indicating regime-dependent winners."
+    measurable: true
+
+experiment_design:
+  research_question: "Which classical forecasting method is most robust in sMAPE across controlled synthetic regimes of trend, seasonality amplitude, and noise?"
+  conditions:
+    - name: "ar_lag12"
+      description: "Autoregressive model with lag order selected from {3,6,12} by AIC on training split."
+    - name: "arima_auto_small"
+      description: "Non-seasonal ARIMA with (p,d,q) searched over small grid p,q in [0,2], d in [0,1], selected by AIC."
+    - name: "sarimax_seasonal12"
+      description: "Seasonal ARIMA/SARIMAX with seasonal period 12 and small grid over (p,d,q)x(P,D,Q), selected by AIC."
+    - name: "ets_additive"
+      description: "Exponential smoothing with additive trend/seasonality (period 12), damped trend optional by AIC."
+    - name: "theta"
+      description: "ThetaModel forecast with default theta decomposition for univariate series."
+  baselines:
+    - "seasonal_naive (forecast y_t = y_{t-12}) as a simple seasonal baseline"
+    - "naive_last (random walk) as a non-seasonal baseline"
+  metrics:
+    - name: "smape"
+      direction: "minimize"
+      description: "Symmetric Mean Absolute Percentage Error on forecast horizon, averaged over seeds and series instances."
+    - name: "mae"
+      direction: "minimize"
+      description: "Mean Absolute Error on forecast horizon."
+    - name: "fit_time_sec"
+      direction: "minimize"
+      description: "Per-method wall-clock fit+forecast runtime in seconds."
+  datasets:
+    - name: "syn_low_noise_high_season"
+      source: "Synthetic monthly series: length 180, period 12, trend slope in [0.0,0.2], season amplitude in [8,12], gaussian noise sigma in [0.5,1.0]."
+    - name: "syn_medium_noise_medium_season"
+      source: "Synthetic monthly series: length 180, period 12, trend slope in [0.1,0.4], season amplitude in [4,8], gaussian noise sigma in [1.5,2.5]."
+    - name: "syn_high_noise_low_season"
+      source: "Synthetic monthly series: length 180, period 12, trend slope in [0.0,0.3], season amplitude in [1,4], gaussian noise sigma in [3.0,4.0]."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 720
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML20.json"

tasks/ml/manifests/ML21.yaml

@@ -0,0 +1,97 @@
+# ============================================================================
+# T21 — Comparing causal structure learning algorithms on small linear-SEM DAGs
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML21
+title: "PC vs GES vs NOTEARS-linear for recovering small Gaussian linear-SEM DAGs"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Causal structure learning methods often claim asymptotic guarantees, but in
+  practical settings researchers usually face small graphs, finite samples, and
+  imperfect hyperparameter choices. For Gaussian linear structural equation
+  models (SEMs), three widely discussed paradigms are constraint-based (PC),
+  score-based (GES), and continuous optimization (NOTEARS-linear). Even when
+  data are generated from the model class these methods assume, finite-sample
+  behavior can differ sharply in edge recovery and orientation errors.
+
+  A compact CPU-scale benchmark can isolate these differences by generating
+  synthetic DAGs with known ground truth over 5-15 nodes, sampling linear-Gaussian
+  observations, and evaluating structural recovery against the true adjacency.
+  This avoids data-download overhead and lets the study vary sample size and
+  graph density in a controlled way. The key metric is Structural Hamming
+  Distance (SHD), which penalizes missing, extra, and wrongly oriented edges.
+
+  A credible study should implement all three algorithms with transparent
+  assumptions: PC via partial-correlation CI tests, GES via greedy score search
+  (e.g., BIC), and NOTEARS-linear via a differentiable acyclicity penalty and
+  L1 regularization. Multiple random seeds and at least two synthetic regimes
+  (sparser/easier vs denser/harder) are needed to reduce variance and avoid
+  over-interpreting one-off runs.
+
+  Because these algorithms expose different tuning knobs (significance level,
+  regularization strength, score penalties), the comparison should include one
+  clear default setting per method plus modest tuning on validation seeds. The
+  resulting analysis should report SHD and at least one precision/recall-style
+  edge metric, then map outcomes back to explicit hypotheses about relative
+  ranking and sample-efficiency.
+
+  *On small Gaussian linear-SEM DAGs, which of PC, GES, and NOTEARS-linear most reliably minimizes structural recovery error under realistic finite-sample budgets?*
+
+hypotheses:
+  - id: H1
+    statement: "At n_samples=1000, NOTEARS-linear achieves lower mean SHD than PC on at least 2 of 3 synthetic DAG regimes (averaged over >=5 random DAG/data seeds per regime)."
+    measurable: true
+  - id: H2
+    statement: "At n_samples=300, GES achieves mean SHD <= PC in all evaluated regimes, and strictly lower SHD in at least 1 regime."
+    measurable: true
+  - id: H3
+    statement: "The best method at n_samples=1000 reduces mean SHD by at least 20% relative to the worst method in at least 2 of 3 regimes."
+    measurable: true
+
+experiment_design:
+  research_question: "For 5-15 node Gaussian linear-SEM DAGs, how do PC, GES, and NOTEARS-linear compare in SHD and edge recovery across sparse/medium/dense synthetic regimes and finite sample sizes?"
+  conditions:
+    - name: "pc_partial_corr"
+      description: "Constraint-based PC using Gaussian partial-correlation conditional independence tests with significance level alpha (default 0.01, optional small grid)."
+    - name: "ges_bic"
+      description: "Score-based greedy equivalence search using Gaussian/BIC score with forward-backward edge operations."
+    - name: "notears_linear_l1"
+      description: "Continuous optimization NOTEARS-linear with squared loss, acyclicity constraint, and L1 sparsity penalty tuned on a small lambda grid."
+    - name: "pc_alpha_tuned"
+      description: "PC variant selecting alpha from {0.005, 0.01, 0.05} by lowest validation-seed SHD, then evaluated on held-out seeds."
+  baselines:
+    - "pc_partial_corr is the classical constraint-based baseline"
+    - "ges_bic is the classical score-based baseline"
+  metrics:
+    - name: "shd"
+      direction: "minimize"
+      description: "Structural Hamming Distance between learned and true DAG adjacency (lower is better), averaged over seeds."
+    - name: "edge_f1"
+      direction: "maximize"
+      description: "F1 score on directed edge existence/orientation against ground truth adjacency."
+    - name: "runtime_sec"
+      direction: "minimize"
+      description: "Per-run wall-clock time in seconds to assess CPU practicality."
+  datasets:
+    - name: "synth_linear_sem_sparse"
+      source: "Synthetic 8-node DAGs generated via Erdos-Renyi with expected degree ~1.5; linear weights sampled from [-2,-0.5] U [0.5,2], Gaussian noise."
+    - name: "synth_linear_sem_medium"
+      source: "Synthetic 12-node DAGs with expected degree ~2.5 using same linear-Gaussian SEM generator."
+    - name: "synth_linear_sem_dense"
+      source: "Synthetic 15-node DAGs with expected degree ~3.5 using same linear-Gaussian SEM generator."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 720
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML21.json"

tasks/ml/manifests/ML22.yaml

@@ -0,0 +1,101 @@
+# ============================================================================
+# T22 — Active learning query strategies for logistic regression on small pools
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML22
+title: "Active learning query strategies for logistic regression on small tabular pools"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Pool-based active learning is attractive when labels are expensive but a
+  modest unlabeled pool is available. For linear probabilistic models such as
+  logistic regression, classic query policies include uncertainty sampling,
+  margin sampling, query-by-committee (QBC), and expected-error reduction
+  style lookahead approximations. These methods are often taught as broadly
+  useful, but on small tabular datasets their gains can be inconsistent due to
+  model misspecification, class imbalance, and high variance from tiny initial
+  labeled sets.
+
+  A useful CPU-bounded benchmark should compare these query strategies under a
+  fixed annotation budget and shared learner, rather than mixing in model
+  architecture changes. The core signal is label efficiency: how quickly test
+  performance improves as labels are acquired. Because final accuracy alone can
+  hide early-budget differences, the study should include both
+  accuracy-at-budget and a budget-curve summary metric such as area under the
+  learning curve.
+
+  The experiment should therefore run multiple seeds, use at least two
+  sklearn-resident tabular classification datasets, and enforce identical
+  initialization, batch size, and stopping budget across strategies. A random
+  query baseline is essential to determine whether sophisticated policies
+  provide real value beyond chance selection. A passive full-data reference is
+  also useful for contextualizing attainable performance ceilings under the
+  same logistic-regression family.
+
+  Practical constraints matter: expected-error reduction can be expensive if
+  naively recomputing retrains for every candidate. A tractable variant can
+  evaluate a capped candidate subset per round and use one-step hypothetical
+  label outcomes, preserving the spirit of expected future loss minimization
+  while staying within single-core runtime limits.
+
+  *Do uncertainty, margin, QBC, and approximate expected-error reduction deliver better label efficiency than random sampling for logistic regression on small tabular pools under a fixed annotation budget?*
+
+hypotheses:
+  - id: H1
+    statement: "At 30% labeling budget, at least 2 of {uncertainty_sampling, margin_sampling, qbc, expected_error_reduction} achieve higher mean test accuracy than random_sampling by ≥1.5 percentage points, averaged over ≥5 seeds."
+    measurable: true
+  - id: H2
+    statement: "expected_error_reduction attains the highest mean area_under_learning_curve (AULC) on at least 1 of 3 datasets, and its AULC is not lower than random_sampling on any evaluated dataset."
+    measurable: true
+  - id: H3
+    statement: "qbc outperforms uncertainty_sampling in mean test accuracy at early budget (10% labels) on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Which active learning query strategy provides the best label efficiency for logistic regression on small tabular pool-based classification tasks under fixed budgets?"
+  conditions:
+    - name: "random_sampling"
+      description: "Baseline pool-based active learning that queries unlabeled points uniformly at random each round."
+    - name: "uncertainty_sampling"
+      description: "Queries samples with highest predictive entropy from current logistic regression model."
+    - name: "margin_sampling"
+      description: "Queries samples with smallest top-2 class probability margin (binary: |p-0.5|)."
+    - name: "qbc"
+      description: "Query-by-committee with 5 bootstrapped logistic regression committee members; selects by vote entropy/disagreement."
+    - name: "expected_error_reduction"
+      description: "Approximate one-step expected-error reduction over a capped candidate subset per round using hypothetical labels and expected future log-loss reduction."
+  baselines:
+    - "random_sampling is the primary active-learning baseline"
+    - "passive_full_data logistic regression reference trained once on all labels for context"
+  metrics:
+    - name: "accuracy_at_budget"
+      direction: "maximize"
+      description: "Test accuracy at 30% labeled budget, averaged over seeds."
+    - name: "aulc"
+      direction: "maximize"
+      description: "Area under the test-accuracy-vs-labeled-fraction curve from initial seed set to 30% budget."
+    - name: "early_accuracy_10pct"
+      direction: "maximize"
+      description: "Test accuracy when 10% of pool labels have been acquired."
+  datasets:
+    - name: "breast_cancer"
+      source: "sklearn.datasets.load_breast_cancer"
+    - name: "wine"
+      source: "sklearn.datasets.load_wine"
+    - name: "digits"
+      source: "sklearn.datasets.load_digits"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 600
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML22.json"

tasks/ml/manifests/ML23.yaml

@@ -0,0 +1,98 @@
+# ============================================================================
+# T23 — Comparing learning-to-rank approaches on synthetic query-document data
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML23
+title: "Pointwise vs pairwise vs listwise learning-to-rank on synthetic query-document benchmarks"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  Learning-to-rank (LTR) objectives are often grouped into pointwise, pairwise,
+  and listwise families, each optimizing a different surrogate of retrieval
+  quality. Pointwise methods reduce ranking to regression/classification on
+  individual query-document pairs; pairwise methods optimize relative order
+  between document pairs; listwise methods directly shape permutations or top-k
+  emphasis. In production IR systems, metric alignment (especially with NDCG)
+  is often cited as a reason to prefer pairwise/listwise objectives, but this
+  claim is difficult to test cleanly on public collections with many confounds.
+
+  A compact synthetic benchmark with known latent relevance structure allows a
+  controlled comparison. If query-specific utility functions generate graded
+  relevance labels (0-4), we can produce query groups with explicit train/test
+  splits and evaluate whether objectives aligned with ranking structure obtain
+  better NDCG@10 than a strong linear pointwise baseline. Because the data are
+  synthetic, we can vary noise and query heterogeneity while keeping runtime low
+  enough for single-core CPU execution.
+
+  A credible study should implement three distinct training paradigms in the
+  same feature space: (1) pointwise linear regression on grades, (2) a
+  RankNet-lite pairwise logistic objective over within-query document pairs,
+  and (3) a ListMLE-lite listwise objective using per-query permutations sorted
+  by true grades. Evaluation should report NDCG@10 as primary, plus at least one
+  secondary ranking metric and a sanity metric such as training loss. Results
+  should be averaged across multiple seeds and at least two synthetic dataset
+  regimes.
+
+  The key scientific goal is not reproducing a specific paper, but testing
+  whether increasing objective-level ranking awareness yields measurable gains
+  in top-k ranking quality under controlled conditions and limited compute.
+  
+  *Do pairwise and listwise lightweight objectives consistently outperform a pointwise linear baseline on NDCG@10 across synthetic query-document datasets with graded relevance?*
+
+hypotheses:
+  - id: H1
+    statement: "ListMLE-lite achieves higher mean NDCG@10 than pointwise linear regression on at least 2 of 3 synthetic datasets, with an absolute improvement of at least 0.03 on each winning dataset (averaged over >=3 seeds)."
+    measurable: true
+  - id: H2
+    statement: "At least one of {RankNet-lite, ListMLE-lite} outperforms pointwise linear regression in mean NDCG@10 on all evaluated datasets."
+    measurable: true
+  - id: H3
+    statement: "The best ranking-aware method (max of RankNet-lite and ListMLE-lite per dataset) improves mean MAP@10 over pointwise linear by at least 0.02 on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Do pairwise/listwise lightweight ranking objectives provide consistent top-k retrieval gains over a pointwise linear baseline on synthetic grouped query-document data?"
+  conditions:
+    - name: "pointwise_linear"
+      description: "Linear model trained pointwise on query-document feature vectors to predict graded relevance via squared error; ranking by predicted score within each query."
+    - name: "ranknet_lite"
+      description: "Pairwise logistic ranking objective on within-query document pairs using a linear scorer; optimized with mini-batch gradient descent in numpy."
+    - name: "listmle_lite"
+      description: "Listwise ListMLE-style objective with a linear scorer, optimizing likelihood of ground-truth relevance-sorted document permutations per query."
+    - name: "pointwise_ridge_tuned"
+      description: "Pointwise linear baseline with L2 regularization tuned over a small grid to ensure a competitive non-ranking-aware baseline."
+  baselines:
+    - "pointwise_linear is the primary baseline"
+    - "pointwise_ridge_tuned is a strengthened linear baseline"
+  metrics:
+    - name: "ndcg_at_10"
+      direction: "maximize"
+      description: "Mean NDCG@10 across test queries; primary metric."
+    - name: "map_at_10"
+      direction: "maximize"
+      description: "Mean Average Precision@10 across test queries using relevance>=1 as relevant."
+    - name: "pairwise_accuracy"
+      direction: "maximize"
+      description: "Fraction of correctly ordered document pairs within test queries (based on graded relevance order)."
+  datasets:
+    - name: "synth_ltr_easy"
+      source: "Synthetic generator: 80 queries x 25 docs/query, 20 dense features, low noise, relevance grades 0-4 from latent linear utility."
+    - name: "synth_ltr_noisy"
+      source: "Synthetic generator: 100 queries x 30 docs/query, 25 features, higher Gaussian noise and feature corruption, grades 0-4."
+    - name: "synth_ltr_sparse"
+      source: "Synthetic generator: 90 queries x 20 docs/query, 40 features with sparsity mask, query-specific weight drift, grades 0-4."
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 480
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML23.json"

tasks/ml/manifests/ML24.yaml

@@ -0,0 +1,98 @@
+# ============================================================================
+# T24 — Benchmarking online learning algorithms under concept drift
+# ----------------------------------------------------------------------------
+# Unlike paper_replication's P01-P07, the "synthesis" here frames a research
+# QUESTION rather than a known paper's method. The model must design the
+# experiment (conditions, metrics, datasets) — we only commit to what a
+# competent study of this topic would include and what the rubric expects.
+# ============================================================================
+
+id: ML24
+title: "Online binary classification under concept drift: SGD, PA, NB, and FTRL"
+arxiv_id: null
+venue: "ARC-Bench 2026"
+paper_asset: null
+
+# The "synthesis" plays the role of the upstream briefing: research question,
+# background, why the question matters, what "a reasonable experiment" looks
+# like. It deliberately does NOT pre-specify a single method to reproduce.
+synthesis: |
+  In streaming classification, data arrive sequentially and model updates must
+  be cheap and immediate. Under concept drift, the relationship between
+  features and labels changes over time, so static train/test evaluation can
+  be misleading. Prequential (test-then-train) evaluation better reflects
+  deployment: each incoming point is first predicted, then used for an update.
+  This setting creates practical trade-offs among stability, adaptability, and
+  computational cost.
+
+  Several lightweight online learners in sklearn-style workflows represent
+  different adaptation biases. SGD logistic regression can track drift via
+  gradient updates but may require careful learning-rate choices. Passive-
+  Aggressive updates can react strongly to mistakes and often adapt quickly to
+  abrupt shifts. Online Naive Bayes is extremely cheap and robust but may lag
+  when feature dependence structure changes. A simple FTRL-style proximal
+  logistic update (implemented with diagonal accumulators) offers adaptive
+  per-feature learning rates and implicit regularization.
+
+  A credible CPU-only benchmark should compare these methods on at least one
+  synthetic abrupt-drift stream and one real sklearn dataset converted to a
+  stream with induced drift (e.g., blockwise class-prior or feature transform
+  shift). It should report prequential accuracy as the primary metric, plus at
+  least one drift-sensitive secondary metric such as post-drift recovery and
+  cumulative log loss. Multi-seed runs are important because stream order and
+  drift-point randomness can materially change outcomes.
+
+  The study should also verify that drift actually hurts models by comparing a
+  no-drift control stream against drifted streams, and should discuss which
+  algorithms recover fastest after drift while maintaining overall accuracy.
+  This makes the benchmark more diagnostic than a single aggregate score.
+
+  *Which lightweight online learner provides the best trade-off between overall prequential accuracy and adaptation speed after concept drift in binary streams?*
+
+hypotheses:
+  - id: H1
+    statement: "On drifted streams, Passive-Aggressive or FTRL achieves higher prequential_accuracy than online Naive Bayes by at least 0.02 absolute on at least 2 of 3 datasets, averaged over >=5 seeds."
+    measurable: true
+  - id: H2
+    statement: "For at least 2 of 3 drifted datasets, the best post-drift recovery accuracy in a fixed window of 200 samples after each drift point is achieved by either Passive-Aggressive or FTRL."
+    measurable: true
+  - id: H3
+    statement: "For SGD logistic and Passive-Aggressive, prequential_accuracy on a drifted version of each stream is at least 0.03 lower than on its matched no-drift control, on at least 2 of 3 datasets."
+    measurable: true
+
+experiment_design:
+  research_question: "Which online method among SGD logistic, Passive-Aggressive, online Naive Bayes, and FTRL-style logistic best balances prequential accuracy and post-drift recovery on binary data streams with concept drift?"
+  conditions:
+    - name: "sgd_logistic"
+      description: "SGDClassifier with log_loss and partial_fit in prequential test-then-train loop."
+    - name: "passive_aggressive"
+      description: "PassiveAggressiveClassifier with partial_fit in the same loop."
+    - name: "online_naive_bayes"
+      description: "GaussianNB updated incrementally via partial_fit."
+    - name: "ftrl_prox_logistic"
+      description: "Custom numpy FTRL-proximal logistic regression with diagonal accumulator and L1/L2 regularization."
+  baselines:
+    - "online_naive_bayes as a simple low-cost baseline"
+    - "sgd_logistic as a standard linear online-learning baseline"
+  metrics:
+    - name: "prequential_accuracy"
+      direction: "maximize"
+      description: "Mean test-then-train accuracy over the full stream, averaged over 5 seeds."
+    - name: "post_drift_recovery_accuracy"
+      direction: "maximize"
+      description: "Accuracy in the first 200 samples after each drift point, averaged across drift events and seeds."
+    - name: "prequential_log_loss"
+      direction: "minimize"
+      description: "Cumulative mean log loss computed prequentially when probabilistic outputs are available (or clipped score-to-prob mapping)."
+  datasets:
+    - name: "synthetic_abrupt_drift"
+      source: "numpy synthesis (piecewise make_classification with coefficient/sign flips at fixed indices)"
+    - name: "breast_cancer_stream"
+      source: "sklearn.datasets.load_breast_cancer transformed into a repeated/shuffled stream with blockwise feature scaling shift"
+    - name: "spambase_like_synthetic"
+      source: "numpy synthesis with sparse informative features and class-prior shift across blocks"
+  compute_requirements:
+    gpu_required: false
+    estimated_wall_clock_sec: 420
+
+rubric_path: "experiments/arc_bench/config/ml/rubrics/ML24.json"